Harpreet: [00:00:06] What's going on, dude, I can't believe it. It is Friday, August 20th already. Time is going by super quick. Welcome to the @TheArtistsOfDataScience Data happy hour superexcited. Have all of you guys here hopefully had a chance to tune into the live session I did earlier this week on Wednesday, did a live session with or mentioned or so. For those of you catching this on the podcast, make sure you head over to my LinkedIn page to catch the interview that I did with or it's also going to be released on the podcast in. In just about a week or so, they'll definitely tune into that. Also, be sure to tune into the episode that I released today with David Benjamin, who's the author of Cracking Complexity, which is a hell of a book. I absolutely love that book. It's free on Audible Rhyme or whatever it's called. The premium subscription of Audible's of you guys have that definitely look for that book because it's a great read. Great. Listen. Um, superexcited have everybody here a shout out to our shout out to all the friends of what's going on, what's going on. Albert Toshie, Eric Marena, Clint Mikiko in the house.

Harpreet: [00:01:16] Spencer, what's going on, Russell? How's it going, man? Superexcited have all of you guys here. I'm ready to get this office or at their happy hour kicked off. Um, let's let's do this, man. I've got a question. I'm wondering how big of a lot know about. But. Planning a Data science project, right? Like, typically, when you're doing some type of project management or planning for some project, you'd like to have these neatly kind of defined timelines, like saying, hey, look, by next Tuesday at four thirty PM, Bob, I'm going to have the insight that you are looking for. But I don't know if, uh, if that is. The [00:02:00] best way to work in Data science, right? I'm wondering if I can get you guys perspective on this when it comes to planning, project planning, project management in Data science, does this kind of predefined waterfall method work or do we need something a little bit more flexible and organic, for lack of a better word, just due to the exploratory nature of the work that we do? Omert, start with then on this one to get his perspective. And then after that, I'd love to hear from from Tom.

Tom: [00:02:34] Yeah, I think

Tom: [00:02:35] So, there's different types of Data science, some Data science acts like a software development project and so forth, shops like that, you can actually run an agile because really it's it's leaning in that direction anyway. So you can run agile, you could run waterfall, you can run whatever you want to. You can make up, you can run docile, you can so you can run whatever the typical software development methodologies are. But then when you start getting into actual research and doing research, you need a totally different process to handle research because it doesn't go in a straight line. It's cyclical any time you start doing iterative data gathering. So even Data engineering can be a cyclical process. And so that doesn't work with things like waterfall. And if you try to slap it into an agile framework, people want to give you these these tasks and items and say, OK, when is this going to be done? Well, I'm on iteration for I'll let you know. And so that's it's a totally different process. And if you need to manage something like research, you have to go in a completely different direction. You have to have a gated process in place where you're not so much managing. When will it be done as much as continually evaluating how much progress is being made and saying, OK, I like not only the amount of progress that's being made, but also the direction it's being made in. And you're making a financial decision to continue funding a whole lot like the academic side works. Where do [00:04:00] we continue to fund this project? And there's a committee that looks at it and says, OK, we're going to reduce funding this year and you really do the same thing. And that's a different type of project management framework. And businesses are slowly starting to understand they need it. But I haven't seen it implemented in a whole lot of shops yet.

Harpreet: [00:04:19] So does that tend to work better? And like so you mentioned research versus software development type of environment. I. The naive question here, but how do we tell which type of Data science team that we're on and how do we tell which which is going to work for the thing that we are attempting to deliver on?

Tom: [00:04:37] Well, it depends on how certain you need to be about the the actual solution, the model that you deliver, the more certain you need to be about it. You either go very, very simple model. That's easy to explain and easy to understand. Or if you go the complex route now, all of a sudden, obviously you're going into the research route because you have to create experiments to validate whatever it is that you've built in that model. So you have the traditional model development lifecycle, but you have this additional validation to the front of it, because the more complex the problems that you that you end up solving for the business or for customers in the more complex solutions that you're forced into, because I'm not a huge fan of going complicated if you don't have to. But when you start getting forced into more complex solutions, if you you need an experimental process. And that's really the cue of OK, from doing experiments and doing research. And that's something different.

Harpreet: [00:05:28] Thank you very much, Ben, for everybody that just joined in, the opening question I left here with was. What type of project planning, project management process makes the most sense for Data science team then was talking about two different flavors of Data science team. You got your research oriented team versus kind of a implementation deployment up of base team. And I really like those insights when thinking so much. Let's go to Tom on this one. And then Mikiko, I thought I saw you here. I'd love to hear from from [00:06:00] you if you are still around. Tom, go for it. And then, by the way, if anybody else has questions on anything, please do let us know right there in the chat whether you're in this room call or joining us on LinkedIn YouTube or which, um, the one person watch which Tom go for it.

Tom: [00:06:20] Yeah, I love the answer. And I just want to take a Tony Stark approach in the first Ironman. Too much to ask for both. In other words, after studying Genesis Burke and listening to his videos demonstrating, it started to occur to me. Wait a minute. But when we're building a machine learning pipeline, we've got all our development mechanisms and we're trying different things, we're cycling through different models, but some of those models are more easy to use for explaining what's going on. It may not be the model we put in production, but that doesn't mean we can't still use that model to explain a lot of our discovery work. For example, I'm always using linear logistic regression. David Langurs always using decision trees. And of course, both of those can explain a lot more, may not put them in production. But I tell you what, this soon as I've scaled my features and I found out what features are most important, I'm running a business that right away, you know, that you can act on this Data here. Now, back to the bigger question. Did Greg and I even designed to talk together on this business and the Data science collaboration and maximize return on the best and the big highlight math Harp we show that.

Tom: [00:07:51] It's really important to take. Your current business priorities in your current Data assets and look for the intersections and [00:08:00] the strongest intersection point is where you have all the data and that's a high priority. But there may be higher business priorities. That's just a clue. Go collect more data assets so you have enough to do so. That's just the thought. Now, my coauthor of the book and writing guide, Supari. Brilliant. I hope you're all following him, guy g h t h. He really borrows a lot of wisdom from our software and software development kubernetes with all their techniques and says we need to leverage from their wisdom. Yes, we do things a bit differently, but that doesn't mean we completely throw out things. We modify what we need. So we're early in this data science age. Take the current wisdom, modify it, don't feel completely stuck to it and use that wisdom that's already there that that's what I got said.

Harpreet: [00:08:58] Thank you very much. The Data gun shout out to Monstercat Silk for giving us the dope beats all day long if you guys do not listen to that streaming channel on YouTube. We got that. It's amazing, Mikiko. What about you? What do you think?

Mikiko: [00:09:15] All right, so I think my answer is going to be, I guess. Kind of a little bit dumb, but in terms of like what process you should use for your project, I would say like look at the culture and the process of the team that you're going into. So if you're doing it individually, so if you're doing it individually and it's like a one off project, you can come up with whatever project management methodology that will work for you. But the minute that you start trying to get that model back into production, you're going to like unless you're like that, kind of like the full stack or the unicorn like Data scientist, engineer, which is not really supportive of the industry, kind of forcing people to be one of everything unless you're one of those. Right. Like at some point you're [00:10:00] going to have to come up, you're going to bump up to like three teams. One is product basically going like, is this a feature or something that is useful, as valuable for the business? Secondly, is engineering. How do you get the model or the Data science assets into production such that it can be used, whether it's internally for like forecast or externally for like a recommendation feature? And the third team you're going to kind of like run into potentially is like a strategy or finance essentially moneybags like who carries that money. So depending on the size of the company, it could be a very lean, small team of just engineers, one PM who is like a product manager, product manager and a project manager. Other times you can have a very fully scaled system.

Mikiko: [00:10:53] Right? So I'd say, number one, it's if you're going into like a big company, try to adopt a system that they have because it's just a much harder fight, especially if you're coming in at a nice junior level. And secondly, like what I've seen is that for some companies where they have like the infrastructure and system, they'll treat it very much like agile. You know, some things you might kind of circle around, like you implement the model and then you decide or figure out it's just not working. Right. So you got to go back to the drawing board. But the company still wants to feature they still want to, like, recommend these products. They just might go about a different way. But like, they have the pipeline infrastructure already set up. Right. But could be a startup where they're kind of building it from scratch. Right. Ground up your number. Ten person at start up in your number one data scientist there, you might just have to be a lot more flexible because you're kind of like building the plane as you're flying in your very seat of your pants. You're you're going to acquire some [00:12:00] tech that you know or whatever. So you're going to have to sort of make it work. So, I mean, that might be my my answer is, you know, wherever you go, especially if it's a bigger, more complex company, kind of figure out what they're doing already, see how you can plug into it. And if you do find areas to optimize, like, you know, just work it in kind of gently and just a science team sport, right?

Harpreet: [00:12:26] Yeah. Yeah. Usually the methodologies are just flexible that you can move. You've got to be fluid and flexible. And then that agility part is a kid like that. I mean, agile, agile in the sense of just moving quick. On down fast, getting up even quicker. I do want to say I feel so cliche, but damn it, um, yeah. So I mean, I'd love to actually hear from, uh, from from Greg on this. I know you just joined Greg, but the question I opened with was, um. We've kind of along the lines of. How do we manage our work as Data scientist, right, like, you know, typical project management likes to have defined scope defined requirements thing, you know, next Tuesday or thirty PM have an insight for us, but just doesn't work like that in in Data science. Right. I wonder what you think.

Tom: [00:13:21] Let me warn Greg that I did mention our talk, but only brief. Yeah, so. I want to make sure I captured the essence of the question, well, it's kind of like how you manage your day to day versus manage your strategic projects. What do you quantify that?

Harpreet: [00:13:46] Well, kind of, I guess more so. How do you how do you go about managing your entire, I guess, work as a data scientist? Right. Like we start up a project and we say, all right, cool. Like in two weeks we want to have this thing, this thing, that thing figured [00:14:00] out and insights delivered and packaged and sealed. Um, but the work that we do is really exploratory by nature. So it's kind of hard to find that balance between wanting to deliver quickly while still. Lowering the Data and making sure you've kind of left no stone unturned.

Tom: [00:14:20] Yeah, so I think you need to capture most of the task inside of a. A tracking tool and. Entitled of the scoping phase, you want to make sure you tackle all the to do list and with that you want to truncate it into deliverables. I think you've answered that partially. Inside of the question itself is that you can never put a finite time for Data science projects because of the exploratory nature of it. So but you have to hold yourself and your team accountable for delivering something. And with that, you have to have deliverables. So what I will call milestones with target dates, and these dates are not finite. They have to move as you evolve. The thing is, you keep tracking your progress towards those deliverables. And if there are obstacles that prevent you from reaching these variables, you update them and throw that tracker. You're surfacing these obstacles and pulling the right team members to help you remove these obstacles and move towards them one by one so they can not be a it should be an iterative process. It should be a dynamic process. There's no set in stone up date that you should set for these projects. At the end of the day, what matters the most is that through scoping, you're capturing all the to do list and you have alignment across the board in [00:16:00] terms of what needs to be done to achieve the project. So in that scoping phase, everyone involved, all stakeholders need to align with it for when there's an obstacle, there is a clear method or racey what I would call responsible, accountable, consulted, informed people who understand when to intervene, to remove these obstacles, to ensure that the progress of that project is ensured is continually.

Harpreet: [00:16:29] Thank you very much, Greg. Um, Tom, you said you wanted to add something here and then after that, I see Russell has some great comments here. And then maybe if, um, if Eric or Monica can take a stab at this question, let me know. If not, then we can continue on to other questions that we got. Got one coming in from Jacob and anybody watching on LinkedIn or on YouTube. If you guys have questions, let me know. I'll add you to the queue. So please do let me know

Tom: [00:16:55] Before it goes to two quick items. The first one is far more important. I forgot that you can play background music while you're on a cell call. So I am listening to Monstercat Soquel. We talk. That's a score, but the second thing, which is just mildly important, I'm going to make a confession here. I'm one of those types that wants to release beauty function super accurate. That's the worst way to go about our work. If we really are delivering some predictions, if our predictions are better than average, like using the average, we're already adding value to get it into production. But make sure you're constantly improving that model. But can't emphasize this enough, and I'm confessing I suck at this, but I've learned get a tracer bullet release crap, call it crap, say I promise to keep improving.

Harpreet: [00:17:56] I like that traceable that terminology from Andy Hunt, a [00:18:00] pragmatic programmer, when their time, Russell, that then Monarchos got some great comments here as well. So let's go to Russell and Monica. Russell, are you?

[00:18:15] Now. Go for it.

Harpreet: [00:18:20] It looks like Russell is having some technical difficulties, Monica. Uh, go for that. You have some great comments here.

Monica: [00:18:26] Sure. I think that it's super important to be transparent to the stakeholders. And realistically, what you can get done in a particular time frame, a lot of you'll run into a lot of problems with Data trying to find the Data connect Data, gather the Data, clean the data and all of that. And a lot of times that isn't known from those end users. They just think that it's just a magic product that they can get in a day or two. So just being upfront and realistic about that and then also super important for my experience and having frequent periodic meetings to show them the progress that you've made and to tell them about any roadblocks that you have.

Harpreet: [00:19:12] Russell, it still looks like incapacitated, but Mikiko has some great comments here at the chatroom. Mikiko, would you mind sharing that with the world?

Mikiko: [00:19:20] Yeah, so I was going to type up another thing to like, I think in the science machine learning, engineering and all the fields that are like super quantum technical, like, I think we as in the royal, we I think we have a really bad habit of basically wanting to promise the moon and then being really, really disappointed internally in ourselves if we don't do it. And if you think about that's like a very like outcome sort of oriented mindset. And it's like exactly like Tom Tom's point, it's actually better to have a model or an analysis or dashboard shit with, like, [00:20:00] your minimum viable thing that works that you can trust and verify and then you can always add on afterwards. But like especially with the business partners, I mean, the reality is that everyone kind of has to chart their career, the course of their career. And trust is such a huge part of it. And so it's one of these things where, like your key stakeholders, like I think I think people are working Data like Data and machine learning for the most part. Like, for example, I've heard people say like, oh, the project manager is not giving us any slack. It's like. It's like how? How early did you have the conversation with them, right, like you don't want to be like the worst case scenario is like you have basically in the process of trying to please people, you've basically told product like, oh, yeah, we totally get that feature out by like for, I don't know, 20, 20.

Mikiko: [00:20:57] And then you go and do the data analysis and you're like, oh man, like we got garbage. We can't do anything with this. And it's not even like if you push a model, it's neutral. It's like if we push them all this will actively hurt the company. Right. Either in the baseline or something that. So you never want to be in that situation. And I think like that part of that is that people wanting to be very service oriented, wanting to provide the best. And also there's competition, too, like, you know, it's a competitive area. So I think it's good to sort of give yourself that slack and understand that, like, look, it doesn't always have to be like you're promising, like, you know, we're going to make the company 20 million more dollars. But maybe what you can sort of promise that you will do your due diligence. You will, like, check all the boxes to ensure that if the product can succeed, it will be a good, viable, resilient thing. It just might not be the most amazing thing. So that was kind of sort [00:22:00] of what I was saying was I like being service oriented, sometimes it's a really good thing, but sometimes it can really hurt you, especially if you're afraid to talk to your business partners.

Harpreet: [00:22:11] Thank you very much, Mikiko, great insights coming in, appreciate all this. Russell, are you back? We will hear you. No worries. Well, we'll circle back to you, Russell, let's, uh, let's keep it moving. Jacob has a question. Let's get to Jacob's question. Um, Russell, have you tried turning it off and then turning it back on? At my word, Jacob was there, the story.

Tom: [00:22:42] Hi, everybody. Yeah, so before we were we working as a design tent with the prop tech startup in the U.K. So I actually resigned in July. So I was thought to probably go on for a while. So the was made a full time little noticed list. And so the challenge actually, because it happens that I do it on a team because my manager and my colleague resigned last month. So they brought in someone who was not too familiar with our Data processes. So are we doing most of the work like getting get some of those are getting Dashboard's and. So part of my responsibility is to public a new design site for the marketing team, so I don't have like. Pryce-Jones and Patrick Opticals, previously I have to be guided or probably told will be told what to do and do X, Y, Z and stuff, so I'm just a bit nervous. So I need like cube, like guide, what I can do, how to scale up on stuff like that physically.

Harpreet: [00:23:54] Tell us an opportunity. I think that's fair. That's awesome. That's an awesome position to be in. [00:24:00] Let's get some tips come in from the start with then. Then go to Eric and then. And then either Monica Mikiko now would love to hear from you guys, but then.

Tom: [00:24:11] I'll do the job, guy, I'll do that guy answer to start out with so everybody can do a better answer after this. So what specifically? I mean, talk to me about the project, what you feel like you don't have in order to do the project that you're tackling. Do you have gotten to the point where you've done a skills assessment on yourself? So as I yesterday, I was trying to dictate, I do like here from our platform for political streaming, then at the Data then create a dashboard. So that's what I've decided to create. And Putsy and Dushka, which I did, I think that particular condition I got for them to retain me. And so I could do Busiek analysis, eSports clinged. It's detering within two or three hotspots. I feel like I'm not like a hundred percent, so I feel like you can be I don't have the core competence. We've been asking the right questions and also so stick with us, basically. What I find with people who are in your position is that you're underestimating yourself. Somebody called a project something different and you're probably feeling like, OK, well, this is actually something different and it isn't. Typically, the business will bring you problems that you've seen before and they'll put different language around it. And it'll make you feel like you're missing skills when you probably aren't. And so my advice is going to be backwards.

Tom: [00:25:48] I would start working a bit on the project and stub your toe a few times and say, OK, I don't know this. I actually don't know this. Then that's going to be a first area to study [00:26:00] and stub your toe a little bit more until I actually don't know this, because what you're going to find is you are way more competent after doing analysis ETL building out the pipelines and doing what you've told me you've done, you're way more competent to take on a project. And it sounds like it's a dashboard project. It sounds like it's a fairly simple analytics project. And so you're probably more capable than you know. So what I would do is start by figuring out what it is that this project requires that you don't have yet. And the only way you're going to do that is to get a few steps into it. And don't be afraid to experiment around and find stuff that doesn't work and begin to create a learning journey based on a little bit more certainty about what you do know and what you're capable of delivering versus what you actually don't know, because you can spend a lot of time learning stuff that you didn't really need. And that's that's the biggest advice I can give you, is just avoid learning stuff you don't need and figure out what it really is that you need to learn.

Harpreet: [00:27:07] Thank you very much. Let's go to Greg and then everybody else with Chime In. Let me go to maybe Entrekin and KeyCorp. Monica.

Tom: [00:27:17] Yeah, I wanted to add on top of what Vin was saying. I think one powerful thing I see people do well early on is understanding where the Data comes from or how the data is generated with regards to the business processes. So if you know which business process you're trying to improve or help, somehow find out the list of your use cases from the business side. Find someone on the business side who can help you understand the Data. I've been interviewed a lot by folks and tables that I'm familiar with and the ask [00:28:00] me how does it make sense in regards to the processes that I manage on the business side and then start documenting what it means for you so you can have your own understanding of these this Data. So what you're doing, you're mapping this data flow with the existing business processes, and by documenting this, you're becoming a powerful person who can now even educate other business folks, educate all the technical folks. So start doing those interviews with key folks who understand what these business processes are and then and then map them out with the origin of that data so you can understand these tables where they coming from. You understand why they're dirty. You understand whether you need to clean them or not and start exploring how this data is stored and so you can have a better way of manipulating them. That that will make you very powerful.

Harpreet: [00:28:58] The troops there, the estimates, the Chicos to the success rate there. Thank you, Greg. Uh, Eric.

Eric: [00:29:11] That stuff, when it comes to whether or not you're asking the right questions, so I just started in my job in my company about two and a half months ago, and I still don't feel like I ask the right questions. I just ask questions. And sometimes that I feel like I ask dumb questions and sometimes I feel like asking questions. I have asked certain people the same questions multiple times because sometimes things are just complicated or hard to remember. You know this. You're only human and you're trying to take in a crap ton of information very quickly, and so I think you just have to make yourself a little bit of grace because you can only learn so much so fast, even if it's even if it's all thrown at you very quickly, you [00:30:00] can't catch it all. And so just ask all the questions, take lots of notes and, you know, and you'll find as you do. The other thing about that other benefit is when you ask the questions, people hear you, people see you. So you've gained visibility and shown like, hey, I'm a person who asks questions. I'm a person, I'm here, I'm real. And so they know that you're there and you become your coworker. You're part of the team. And that's an important the important thing that's important piece to build. It is not because, you know, not for a selfish reason, but because, like, they need to know, like you are part of the herd or the pack or whatever aggregation of animals you prefer to be like and just know that know that you're there and a part of their.

Tom: [00:30:45] All right, thanks, Eric.

Harpreet: [00:30:47] Let's first hear from you.

Mikiko: [00:30:52] If you want to also find out fun names for groups, groupings of animals, you should look that up

Tom: [00:30:58] Because they're like

Mikiko: [00:31:00] Atila of gators and there's some fun stuff there. Yeah, something I do. So I guess two things. I remember when I had first like like moved to finance, I was working as an analyst there and I was terrified because I was surrounded by all these MBA types. And it was just the culture shock was real because these were all Wall Street types. They were in suits. I came in with sweat pants and like a button down because I was running late. One of them was like, aren't you a bit junior to be like doing casual Thursdays? I was like. But two things that I think for me, I notice like so one, for example, like when you ask questions, sometimes it can kind of feel like if you're asking a question, it feels very elementary. So I would sort of like rephrase the question. So, for example, there is a number of different ways to calculate, like revenue or annual revenue. So instead of asking them, like, what does annual revenue mean, I would go, how is annual revenue [00:32:00] calculated here? Like, what are the fields or Data? What is a field or used? And I could say, because when I have seen it in other places, this is how it is calculated, I want to know what the mapping is, the Data. Right. So that's one way to kind of phrase where it's like it's like I know the concept. I just there's this specific part of it that I'm missing because people would use the same name for, like, different calculations. And I remember when I was first going into finance, I was like, oh, yeah, there's one calculate.

Mikiko: [00:32:28] There's like one definition for everything. And there might be one definition, but there is actually like a million implementations based off of where you go. So that's one thing I would do is I just try to phrase the question in a way. So I'm still asking the question, but it kind of makes me feel a little bit better. But I'm still asking it. That's the important part, is to still ask the question. And then I think the second part, what I would do is when I'm taking notes, you know, like first of all, like in three parts, all the facts, all the sort of assumptions and then all the unknowns. Because what that does is it empties out your brain because sometimes you can be stuck in like paralysis and it's like, oh my God, like, you know. So like, for example, like, what does it all mean? Like, what do they want with it? What did it do? And then what you can do is you can say like OK, like the known is they want Data to get transformed and dumped someplace else. The assumption, which may not be correct and that's something to sort of confirm is I need to pull Data from place to place. Right, so it's like you need to go from from one location to another, but what they might actually mean, right, is like, oh, you can write your own scripts to clean the Data in the instance. Right. So maybe in the dashboard, instead of writing any script, you can find a way to write a macro within it so that it just recalculates values.

Mikiko: [00:33:57] So but it's good to call out like what were the facts like, [00:34:00] what are the assumptions and then what are the unknowns. And then that way, for example, when you approach people and you're like, can I can you feel this like fill this in for me? Right. You can also confirm, like, are these like these assumptions. Correct. And based off those unknowns, like like LinkedIn said, like it does help to start working with the project because then you can figure out what is the right sort of like learning resources you need. Right. Because for Etel, if you look it up, for example, a lot of people say, well, it IBM or whatever tool technology, but maybe it's actually for your particular project. You might not even need to know all that. Maybe you just need to know like where how to write a script and where to host it. Or or something like that. So that for me, really helps us write notes and outlining specifically what it is, I don't know, so I can go back and fill in the question. And then also when I ask people like, you know, can you you know, can you define this for me? Can you do X, Y, Z for me at all times? Business people, they'll say stuff and then they'll kind of assume that, you know, exactly like what they're talking about. And sometimes it's good to call them on that and go like actually we are we do not have the same definition or the same thing in mind.

Harpreet: [00:35:13] Yeah, that was huge for me when I was at bowled commerce, I had to like they're just throwing around a lot of words and a lot of terms. And like, I was new to e commerce at that, you know, at that time. And I had, like, no clue what it meant. But then I was like, it doesn't mean I'm stupid to ask questions. Like, I just be like, oh, you keep saying these like words. And I hear them. They don't mean anything to me. Please, just real quickly, break it down. Tell me what this thing means and and what this means for the bottom line of the business. So I can have some understanding. So it's completely OK to ask questions. No questions. It wasn't intended to be done, but Monica, go for it.

Monica: [00:35:55] Just one other thing to add is just I'm all about transparency, so don't be afraid to say [00:36:00] know you don't understand something and to ask questions, something that I noticed in my audit days, they would feel kind of like attacked or something. If you ask them, like, how does this business project work or don't give me the definition of something, they would kind of be hesitant because they didn't want to say the wrong thing. So if you were to like Mikiko saying, rephrase the question, like how do you view this business process? What do you think that would be helpful for me to know about such and such? So. And there everyone hit it on the nose, so

Harpreet: [00:36:43] Thank you very much, Monica VIMS got a comment here, assertively ignorant. When I get jargon, I start asking questions with confidence about how little I know. Sound strange. It works well. Yes, it does. I was in a meeting with some lawyers and accountants and they just kept saying stuff at the possum's like yo yo motherfuckers keep again with these words. I don't know what they mean, like, but talking to each other and talk to me and let me know what the hell you are talking about. All this money and talk to me. Um, yeah. So anyways, I take away from that that man, anybody else got any tips for for for Data shout to Ben Taylor in the house? But I've been Taylor.

Tom: [00:37:22] That's how I think I feel sorry.

Harpreet: [00:37:24] Good man, you are awesome, man. No, no, no explanation for why you feel that way, because you are. Keep going.

Tom: [00:37:33] I got to say real quick, Ben Taylor, you were freakin awesome today on Lights On Data not really enjoyed listening to you a time. Thanks, man.

Harpreet: [00:37:46] Did I miss that I didn't know you were on late Sunday this morning. How'd that go?

Tom: [00:37:50] Oh, we were talking about meaning meaning of life and all sorts.

Harpreet: [00:37:53] Just turning into a treatment episode

Tom: [00:37:57] Like, yeah, I'm like. Yeah, [00:38:00] I mean, not nice to your kids, it's I guess that's a profound statement. I'm just kidding.

Harpreet: [00:38:04] I didn't listen to a lot of Lex Friedman lately. It's been interesting. Um, didn't really, really doubling down on this. Uh, I've been really thinking hard about well, not really thinking hard. Just I have been thinking about ethics and stuff like that. I've been reading this book from Max Tegmark or Life 3.0. Oh. Uh, I don't know if you guys have checked it out yet or not, but I highly recommend it. Or at least just listen to, like, the talks he gives around the book or listen to the talk that he did with them. Lex Friedman, he paints a very interesting picture of what the world could look like in just our lives lifetimes in just the next few decades with all the advances that we're making. Um. So if anybody has questions, let me know, uh, I'm keeping an eye out on every war, on LinkedIn, on YouTube and in the chat itself, and I don't see to

Tom: [00:38:55] Want to have one little thing from earlier. Greg and I, when we give our little talk together, we've been trained, we lose various we've been trying to emphasize, but we've got this. And by the way, I can see that not everything is a clean machine learning development pipeline in the work we do. I certainly haven't been doing anything. I clean the last two years, but we talk about it like the basics is the foundations, but. You look the queen. Anything else you're doing something like this, if 80 percent or more of our work is getting ready for modeling, getting the Data ready for modeling, there's so much value we're discovering in that process. There's many. Stepping stones along that process where there's some key Data we can go communicate back, I use the illustration of the feature weights, for example, but there's other things along the way. And to me, [00:40:00] one of the bigger things is you discover dirty Data don't just write the cleaning routine for it. Go back to the Data management system programers or the Data collectors and say, would you please not allow an old value here? Would you please make sure this is type checked? But little things like that just save enormous amounts of time later.

Harpreet: [00:40:23] Thank you very much, Tom Eric, with a very good observation. Meetings must have just ended in Mountain and Pacific Time because a bunch of cool folks just got out into the chat. Shout out to the cool folks who just joined Ben. What's up, Mark? What's up, J. What's going on? Super happy to have all of you guys here. Um. What you guys take this next man, I have no questions coming out of a question I'll go for. Yes, please. Yes.

Tom: [00:40:47] So do you guys think a. I guess I don't know if there's a term for this, a good or good data scientist can become a good startup founder. What does it take someone heavily in the research and did a science get this idea and says, hey, I want to oh, Ben is interested in this question, like, what are going to be a businessman now? We'll get to the science, but let's do this.

Harpreet: [00:41:19] I'm interested to hear the benefits of a successful entrepreneur with an exit as well. So go for it then.

Tom: [00:41:25] I, I think it's Data scientist. It'll sound like I'm generalizing, so if you're a data scientist, you're pretty smart, like you're stemming a lot of things, your multi discipline. And I think sometimes that can get you in trouble because you go into a startup and sometimes people focus too much on the tech. They're like, we're going to go build this innovation. Deep learning is going to go faster. We're going to do this next week. Things can be better. It's be faster. And I think the conversation that's missing usually is who's going to pay you and how much are they going to pay you in? Just because something is cool or fast or interesting, [00:42:00] it doesn't say they're going to pay you. And so with our startup, I remember initially I was embarrassed when we get to pricing, which is hilarious because we should actually have a conversation like why would I be embarrassed when I'm asking you to pay me forty thousand year, eighty thousand a year or two thousand a year? It's because I'm not convinced of the value. I'm saying, hey, we've got the steepening platform. You should really consider it. You can do some cool models, you can do some stuff. And so I think the the word so. The answer to that question, Greg, is no and yes, it's it's a matter of how quickly you can get to value. And so if you care more about value than your customer, then you'll get there very quickly. If you fall in love with a tech and you forget value, then you will definitely fail. And I'm not speaking like I have all the answers, but I've definitely had my ass kicked. So that that's the word of warning is get to value, not the value from the Data sense perspective, but the value from the layman buyer who is willing to cut you checks but cut you checks where they don't get fired, cut your checks where they get promoted. That definitely makes sense. I do have a follow up question, but definitely would like to hear other people on this in terms of startup, so

Harpreet: [00:43:14] Get to do an open question. Otherwise we could hear from Mark or Dan on this as well. Up.

Tom: [00:43:20] Well, I wanted to hear hear the other responses first. OK, great.

Harpreet: [00:43:24] Let's go to you to Mark.

Mark: [00:43:28] So definitely I don't consider myself like a full on founder, but I've tried a few times and what I've realized is Data science has not helped me be better at business. If anything, business has helped me be better at Data science. And the reason being is like when I'm when I'm like going through the first iteration and like, all right, what's our business model? What's what's our town like, what's what how can I figure out these pieces and [00:44:00] drive kind of like this. This is something needs to happen now. The market many times when I don't have access to Data even work with to do those things is coming from a lot of user interviews. And that's like really qualitative Data. And many times, like, if you go straight to the Data, like I don't even know if I need that any build infrastructure to even work with all that Data. And like my time is better spent talking to potential customers and talking to potential investors and thinking through this business idea and like work that's in the market and the Data can come later. But because I spent a few times going through that iteration of my own ideas or for other startups, now when I go into my Data science work, I'm like, oh, shoot. Like, these things don't matter what's going to drive the needle forward to get, like, their first customer or like get some key customers and how it fits in this market and how does my insight fit within that? So I think Data science does not make me better at pursuing my pursuing my passion for for for startups. But it does work in the reverse where it's definitely helped me as a data scientist in the startup space.

Harpreet: [00:45:14] Thank you, ma'am. Let's go to a Mikiko.

Mikiko: [00:45:18] So I tried founding a startup that was amale based. And I don't want to say it didn't work, but I am not a part of that venture anymore, and I think so here's kind of similar to what Mark was saying. I don't think working Data science hurt me. I think us sort of not having a strong, committed technical founder, I think that sort of hurt us more frankly, because I think so the story and this person was sort of like you [00:46:00] start hopping on that project was the story was good. The positioning was fantastic. The product very well needed. One hundred percent understood like the segment of the market. But I think, like not having, like, a strong, like, engineering co-founder that could help us with privacy, security and also just infrastructure like the devil stuff that really, really hurt. And I think part of it was like a little bit of arrogance. I assume that because I worked as a data scientist, I did some code that I could figure out, OK, the entire infrastructure, and that just did not work. But I don't. So I don't think working as a data scientist like Hertz. Right. Because I think especially if you're trying to push like an email product. So, for example, Josh, Toby, Josh, Toby, and like they each surgey who teaches feels like he, I think co-founded or he founded Churton Dotcom.

Mikiko: [00:46:56] And they use eye to detect whether or not you are plagiarizing. Right. And so and Josh was, I think, working on something right now. It might be his first venture. So having that kind of language in that domain, understanding it does really help, I think. But I feel like the reason why a lot of email startups probably don't work out or it seems like they're working and then poof, like, you know, you find out later on that like, OK, 20 million dollars of funding just went nowhere is a huge part because they're not getting the funding like there was in Canada. I think there was a pretty well known one that was like, what, within the last year or two that that happened. But it's not they're not getting the fundamentals right. And I think there's this really fun quote from Edith Wharton. Who said all happy it was like all happy families are alike in the same ways, all unhappy families are different in different ways. And I think like, you know, stars who do really well, there's a lot of things that have to go right. And also there has to be that opportunity. And I think when stars fail, [00:48:00] there could be a number of reasons why each one feels right.

Mikiko: [00:48:03] They didn't get market fit. They didn't develop the engineering. They didn't do X, Y, Z there. They were doing something that ended up looking very similar to a patent that Apple and Google holds. And they got sued out the water. So it's really they just didn't cover their bases. So I feel like as long as you have your foundation, your bases covered and it doesn't have to be in the same co-founder. This is why law because they like to like people who have to write. If someone dies, then the company can still go on. And I do know obviously where that happened to their company, like someone died, but they still have the other co-founder to keep it going. But part of it is also because different components of that team will bring sort of different skills and experiences and insight. So, I mean, that's that's just my two cents. Like, I think I would love to do it again in the future, but I think next time I'm like, you know, now that I know how we get the story and the pricing and all that, it's it's about kind of like the foundation. You can't have a Emelle if you just don't even have, like, a good engineering infrastructure for, like, the regular web mobile stuff. So.

Tom: [00:49:08] In Mexico, you should all quit your job and go to a startup. You will learn something, I loved everything that what you were saying, Allegri. And aside from saying what Mikiko said, just elaborating a little bit on a couple of our points, when you're when you found a business, especially a startup, you're peeling back an onion of problems and Data scientists tend to be tacticians, which is problematic. We want to do stuff. We want to build stuff. We want to ship to customer. We want outcomes, outcomes, outcomes, outcomes every day. And running a startup isn't like that. Running a startup is consistently slamming your head into a new problem every day, finding [00:50:00] someone who can help you solve that problem or sometimes learning an entirely new skill set to solve that problem. And a Data scientist is kind of like a salesperson as a founder. When you get a salesperson as a founder, they think, OK, we've got a sales problem, OK, every problem is a sales problem. We just have to sell harder. We've got to sell better. We need a little bit of change the product and we'll be able to sell data. Scientists is going to look at it and go, well, OK, so it's a data science problem. And it's and you see it from that kind of one dimensional world view of the thing that you are. So in just day in and day out, that's what you do.

Tom: [00:50:38] And it's the same thing with salespeople. They say this is what they do. And so when they run startups, they tend to look at every problem as a sales problem. Data scientists tend to do the same thing. So like I said, you're peeling back the problems. And to be a successful founder, you have to be a data scientist who doesn't mind solving a sales problem. You have to be a data scientist who can recognize that this is a product fit problem, that this is how I solved my problem with this. And if you look at all the Mellops tools that are out there, you know, I solved my problem with my product. Now I have to talk to customers and solve theirs in order for this to really grow. And that's another type of problem. And it's not a Data science problem that's getting with customers and understanding. But the thing is, you can use Data science to solve so many of these problems that we really get kind of pigeonholed into thinking, OK, it's just another Data science problem. I just got to get some more data and I'll figure it out. And as a founder, if you don't have a good team around you that can pull you back, that you respect and trust, who can pull you back and say, look, then it's not as a science problem. All right. We're going to do this with an algorithm in a massive amount of data gathering it.

Tom: [00:51:53] No, I'm just going to send a few of our people out to talk to our customers. So we're going to figure this out. And so you really have to have [00:52:00] a good team around to do the scientists. But the second side is you have to have a data scientist who's willing to change and not use the hammer every day. Thank you. If I can ask for a minute. Hey, guys, this is my first meeting actually. So prior to this, actually, I ran my family business, so I was a small business, so I wasn't a startup. So I came to Data science from an entirely different perspective. I came from, OK, I'm running a fifty million dollar business. Technology is not a priority at all. So you have technology which you use sometimes, but we are not investing in what's the latest technology? How are we driving this technology? What knowledge that we're getting we're getting is what's driving that the most to our value. And in most cases, I came from a quantitative background. That's what I did. And it was dealing with people selling, dealing with lawyers, really business strategy. How are we improving our business model, our operating model connectivity? The top of the stuff that's been talked about, a lot of a lot of the is the focus is how we're putting off operating model model.

Tom: [00:53:22] How we drive value to our business is not really what technology technology comes after, after we have all these processes, after we have all the people in place because we don't have the people and we don't have customers, technologists at least to take a small business that has 50 million dollars in revenue. There is it'll it won't work. It just doesn't work. It'll fail. So that's from my perspective. Thank you, gentlemen. Yeah, thank you. Go for it. Yeah, that that's that's awesome. I really love the discussion. I guess the follow up question [00:54:00] I have is that I was following an episode from Jason Calacanis and he's like using the vaccine. And in terms of what you should expect or what VC firms, the legacy ones, especially what they expect is startups to come up with a team where there needs to be. And maybe Mikiko alluded to that there needs to be someone who's strong in the technology side, because if you show up by yourself and you show up as the business person, they might say, well, you're not technical, technically savvy. So we're not really interested in investing in your company, I guess is that is not the right mindset that VCs should have or do we really need a co-founder who's tech and a co-founder was known tech to have a solid, solid startup that has a chance to survive the next phase of obstacles?

Harpreet: [00:54:59] Someone's got to build it. Someone's got to sell it, right? So, uh, Mikiko, I see you had your hand up over it.

Mikiko: [00:55:05] Yes, so as a result of basically not getting an engineering oriented founder, we had to engage contractor services which were spotty cost money and equity. And when that project ends, that's when you find all the regressions in your code and also to security and privacy is a new thing. So I think it's one of these things where, like. I mean, there's all things you can say about BK's, right, some good probably a lot of bad, but at the end of the day, right there, there are also business people. So they just want to see a return on their investment. So they're going to just basically try and mitigate that risk as much as possible. And that's like the unfortunate thing is that it's it's known that if you're, for example, a woman, if [00:56:00] you're LGBTQ, if you're a person of color in all that. Right. There are certain things about founders that will sometimes decrease their chances, which is incredibly unfortunate. And the nice thing is that there are now like VXI funds that are specifically so start out. For example, they're very focused on LGBTQ then and there's like other female ones too. But I feel like they are just going to mitigate mitigate risk as much as possible. And so if you're selling like an engineering product and you don't have like nothing, say, like an engine co. Right. Even if you have a number three employees, like an engineer.

Mikiko: [00:56:37] Right. As long as you can basically say, like I have the engineering resources where we can build the project and all that, that will go a long way towards that risk mitigation. But there's other ways, too, that they could be a little bit concerned. So, for example, you could have two engineers who just have no idea how to talk to people, no idea how to market, no idea the value prop. They show up in a meeting and the voices are pitching them all these questions. And the voices are going like, oh, man. Like they're just saying yes to everything. Like, you know, we're getting them for we're getting them on the equity, like, you know, what else are they going to say? Yes. Are they going to, like, survive in the marketplace? Like, do they know the domain? So there's a lot of these ways that look kind of similar to an interview. Right. Like the VCs can kind of pick out like, OK, well, we don't like this about this. We don't like this about that person. We think they should have this in this. Right. So as one of those things where you just kind of have to cover a lot of those core bases as much as possible in some way. And then at that point, I think it's kind of up to like the back up to like, look an opportunity to some degree.

Harpreet: [00:57:42] But Marc, Ben, Ben, Ben, Tom.

Mark: [00:57:47] Mexico's great point also reminded me something that came up when I was speaking to angel investors and other other businesses or for idea like the biggest thing that came up for me and my co-founder at that time was like, you're both first time co-founder [00:58:00] co-founders. And that's a huge risk for us to try to pursue that. And so the way we mitigated that was that we spent weeks going through our network and trying to find some who is open to to join us. And I was also already a founder and had VC backing already before. And so once we add that person into the mix, when they have the expertize of like how to navigate and also had connections within the space, but also like like reduce the level of hesitancy to give money to people who are completely new to this. And so that was another key thing is like are you a first time founder? And so if you are like the advice I receive, it's like partnering with some who's actually gone through this process before or at least getting a set of advisors that can kind of reduce your risk to two potential investors.

Harpreet: [00:58:50] And then after Ben, we'll go to Tom.

Tom: [00:58:54] Tom was ahead of me. Oh, for. Ben, thank you. That's perfect because I kind of wanted to end this question of you of. I had to start a I'm not going to say it's. It's just been on the shelf for maybe about a decade. But a question regarding this whole thing, we all think it's conceivable or maybe preferable to make your initial startup or at least one of the 10 initial startups you try one that you can just launch by your own cash flow by trying to follow both good tech and business practices because. I've been part of other startups and I've seen all the classic mistakes that have been listed here today. I'll give an example, the most prevalent one, the most salient one from the things we've been talking about today. This, [01:00:00] the guy that was leading the startup, could not get his head out of the hardware, out of the tech and seemed to just be making boneheaded, stupid business moves and not leaving the team very efficiently because he was always focused on the tech. But when I've talked to successful startup people, they've always affirmed when I've asked them. Would you prefer to try to start where you're you're making it on your own cash flow instead of having to get investors, it's like. No pause. Yes. And I'm just curious how many of you think. It's healthier to try to start that way, because if if it's pie in the sky, thinking, I'd like you all to shoot me down, because I actually think if you get to that level of business and tech ingenuity, it's so much better to launch that way.

Tom: [01:01:06] I mean, if you don't mind answering first, that would. So with our startup, we had a lot of pressure from the CFO of HireVue, who was my friend, saying try to hold onto the equity, try not to raise like try to get to a million revenue, 10 million revenue and. I'm a big fan of racing, so if I went to another startup, I'd raise as much as I could of the gate rather than sell funding, like maybe I'd get like an MVP or something or first customer if I could. But growth capital does so much. Just the talent you can hire, the the mistakes you can make. Data robot. We've raised a billion dollars. You can do a lot if a billion dollars. I'm not saying that that's a rare event, but raising any capital can be really useful. What when one of the concerns [01:02:00] I had Greg with me when I was doing or not. Greg. I'm pregnant, as I have with me is I always had these side projects, like I had the super technical thing I was working on, and if you can raise capital, you can go full time, you can quit your other jobs. You can actually hire competent people. Probably people on this call come join you and go after it. So I but people have different opinions. My co-founder you would not raise again. I would raise again. So you're always going to get two sides of that coin when one thing I wanted to bring up with this group is their their valuable lessons for criticism during the startup phase.

Tom: [01:02:37] And so great. One of the lessons I learned is if you came to me and you're doing a startup and you said, hey, I landed my second customer. That should be a point of celebration, but what I've learned is, depending on the details of your second customer, I could criticize you. I can say the second customers, two different from the first. And you could try to argue and say, no, there's a platform play or the technology is being shared. But a fair criticism would be. But the deck you use to sell the second customer had nothing to do with the first customer. And the momentum you gain from the second customer did not earn you to the first customer or second customer. And so there is so much value when you do a startup and you come with an idea, get those three customers in a line. They don't have to be big accounts. But just the fact that the first customers helping you sell the second and the third, that is something that VCs talk about where when I was young founder that I was so arrogant. Platform play. We support these Data types and we sell to everyone. And that is a death knell to investors for good reason. I didn't understand that early on. This is such a good point then, because I feel like the way you're answering, you're seeing too many startups slip into the specialist quadrant of cash flow quadrants instead of the business clock. What a kind of egregious thing. And the other thing I'm hearing, too, is that this are looking for some sort of morality score or product, right? So this [01:04:00] brought this customer spreads the word and it amplifies throughout.

Tom: [01:04:05] Thank you, Ben. Is that the magic thing for them is can you get to three to five customers? And if you can, because the issue that comes up that data scientist can trick themselves into is just kind of insulting. I say it's just insulting because if you actually went in charge your consulting, you'd be making more than running after the startup of this platform and building this thing, this technology. If I look to your hourly rate and what you're doing over here, it ends up being just kind of insulting. And so sometimes you can trick yourself into doing all of this work and the momentum is not there. And Gray can only be scaled because it really it comes down to your scale factor, like what is Greg's scale factor if you're working one hundred hour weeks to deliver across three to five accounts? The sad reality could be if you step back, you're doing discounter consulting. And that's where you get that check, you really check yourself mentally and say, how do you how does the second and third customer become easier and how do you actually most founders don't take vacation. And that's a really sad thing. That's a sad reality. How do you take vacation? That should be the first milestone, Craigslist take a vacation in six months of the startup, then we'd all celebrate on this call like Greg did, something to scale in scaling is hard. Ofra Mikiko.

Mikiko: [01:05:22] Again, it's to reiterate, sort of like Ben's points right up your so it's so so the start was passed. So I'm doing something different now on the side that is a lot more manufacturing and distribution having. It's not technology. Right. But I think one point is in terms of like do you raise so for example, MailChimp, whose opinions I do not represent and safe harbor illegal, yada, yada, yada. I don't speak for them. Right. But they're a private company and they've been private for 20 years and they don't do any B.S. funding. And a lot of times when they're asked, why have you not? [01:06:00] Why haven't they said, well, because we started MailChimp during the dotcom boom or sorry, dotcom bust, boom bust and nobody was willing to fund it, so they kept it as a side thing and they kind of did consulting whatever to pay the bills. And eventually it made more money than they did and did so the whole spiel. But for them right there, they don't have easy funding because they just it was not available to them. And as a result, it has created very in some ways unique company where they're not tied to the same sort of goals of every quarter. We have to post profits. Well, we don't because we don't release them publicly. But they still made six hundred million dollars in revenue last year. And that I can talk about because they did talk about that in a blog post. Right. For me, if I was to start being very similar to Ben, I think I would probably raise early as hard as possible.

Mikiko: [01:06:53] I would I would get an MVP as soon as possible and talk about tech that screw that. Right. Like, don't don't even worry about tech debt. Like build something, get in front of people know. And that was one I think one that was one of the things that killed us was that we were trying to go for like perfection, robustness. And I'm like, we should just create like a bunch of like. Not even developed as much as we should have just taken off the shelf models like Roberta's or something like that, you use an MVP but but showed how it could be solved like a pain point. And then we should have raised to get like the security and the engineering and developed resources that we needed. Because the other part series, if you can raise and there's a lot of research out there, you can look at a 16, which is the end. Anderson Horowitz website. There's a few others that you can look at. But if you raise like I think like a certain level of funding in, like, your seed stage. ET cetera, et cetera, right? That's like a really good signal to engineers or to designers or to marketers or growth people that like, oh, like this [01:08:00] staff knows what they're doing. So a lot of times those employees are willing to join if they get more equity and less cash. Right. They might not do sweat equity. A lot of times that's like KOF that's like the co-founders of Sweat Equity.

Mikiko: [01:08:12] But if you can raise a certain amount early on, that already is a signal. So in a way like you can get that talent and you can build up that infrastructure and then you can just focus on like how do we get from zero to one and then how do we get from one to like one hundred a thousand all that. And it just it really depends also on your favorite flavor profile. Now, the other thing, too, that I would probably do differently, honestly, is that it is such a hard road. It is so hard. Like I was working on it for about a year and this is after having worked at a bunch of early stage startups. So I thought I knew my shit right because I had been in ICI. I'd never been the founder position, but I worked in early stage startups, so I thought I knew what I was doing. But I was working on it for a year as a side. Right? So in a way, I was tricking myself into thinking like, oh yeah, it's going well and it's going well because I had a full time job, so I was paying my rent and all that other stuff. So when I quit my job to go try to get it over the finish line for six months. That's when the panic started and I was like, oh, man, like we're not making the progress on the roadmap that we should be, and then all these other issues start popping up and all that.

Mikiko: [01:09:23] It's like hell with work from home. You saw a bunch of couples like divorcing these now, like both the husband or wife or or wife and wife and my husband were at home with the kids and they're like, oh, we actually don't like each other all that much. And you kind of got that like from work. And it was very, very similar work on Staab. Like, if you are working on a part time, you could trick yourself into thinking that it's going places. So on the one hand, you keep your job, you can you can mitigate risk. But also if you are not sort of immersing yourself in it, you might not really be seeing kind of like the red flags and the blind spots that ultimately get [01:10:00] you. And I six months after that made the decision to I'm going to go get a full time job and I need medical insurance, any money. But next time I probably wouldn't do it unless I was, like, really passionate about the product, you know, and it doesn't have to be like it's world saving. But I'm like, next time I got to I got to love this shit. I got to be really passionate about it because it's going to be hard no matter even if you do everything. Yeah. Even if you do everything right, it's going to be hard, you know, but I still think it's worth it. It's worth it once in your life to try it.

Harpreet: [01:10:32] Then go for it and then go to

Tom: [01:10:34] Mikiko says something to me, and that's sometimes we overengineer like we're working on a startup, we're overthinking it. And I remember with my co-founder and model, the controllers came out with SQL. I started back in 2003. So for me, that was very confusing and like, oh, man, these Monaldi controllers, I just want to write my sequel. And I had an opportunity to talk to Eric Reese. He wrote the Lean Startup, and I asked him about this, about the technical dead of writing outroar SQL queries. And he said, you know what, the funny thing about technical debt, especially in a startup sense, sometimes that debt never comes due to the important point that they're screaming right there with technical debt in a startup, you should never overengineer anything because you have no validation the fact that a single customer is using something. So my recommendation to people is don't just do an MVP, do like an MVP, do a shitty MVP. And so if you're building it out, if you can't tell me, like, yes, if you sneeze, if there is an update, if you do anything, this will break. But currently today for this demo or this customer, this works. You can land the customer. You can you can improve. You can actually react to a customer or therapist to you because the product's networking can jump in and fix it. But I think too many founders, they want to like to kubernetes out of scaling like that, you know, all these technologies that you should be using. And the thing that is so sad about that is you don't have a thousand customers. You're fighting for three supplicate that's like that is like the alarm bells going off, like you're fighting for three customers, I know there's all these sexy technologies. [01:12:00] The technical debt may not come to celebrate technical debt early on.

[01:12:07] Get rid of it later.

Harpreet: [01:12:11] I think I might be starting to start up after this with all this advice, but go for it

Mark: [01:12:17] Again, Mickey, awesome advice, inspired of a five mind and just talking about how how much being a founder like it's really glamorized online, but like, it really sucks. And again, some of the things that start ups and like I've been around friends have had startups. I'm like, they're totally doing I can do it, too. And when we're applying to Y Combinator, Parvizi, you get one hundred thousand dollars for your service seed round, essentially. And when we've gone to the final round for interviews, it's going to be like the no go out to interview and leading up to interview. I was both excited but also having like panic attacks because I knew that if I got the money, shit hit the fan. Not in the sense of like, well, yeah, everything, because the start of everything is on fire, but because then it'll be real. I took money and I have to deliver and there's only one hundred K for our business and our three co-founders. So I'll be the living. I'm not to convince my wife that we're going to move back into my parents house so we could like, actually pursue this business and basically drop everything and like that. I was passionate about the idea. So like I was willing to take that sacrifice, but. I think a lot of people don't realize the sacrifice it requires to actually do that because they have like the founder's salary. And, you know, I think a lot of people don't take that account as well as I can.

Mark: [01:13:37] You can you handle six months of just having nonexistent pay or vacation, anything like that. And that was like a profile to that. I love like when we're talking to angel investors, like one of the things they're asking, like, how hungry are you? Are you willing just to work 80, 90 hours? Because it's not I'm not I don't want to give you money. They should have just told us that, which is completely fair. And [01:14:00] so, you know, that's just something that goes market goes up just like also for the founder profiles. Are you willing to do an ultimate decision of like after you get into Y Combinator? We we still had traction for potential customers. But because our product was selling to pharmacists, it was like during the eight to five was like, I can't work full time. I'm at an inflection point. You guys go all this or it's not going to work at all. And I just I couldn't afford to to be a founder. It just didn't work out. And that was a tough pill to swallow for myself. And it really puts perspective. I put in like six months of work into this, only to be like, oh, I can't afford to pursue this. And so now my head, I'm like, how can I better prepare myself to make that jump as the founder?

Harpreet: [01:14:51] But I've got to go to Marina after this, that but I mean, just to quote Novara, because he talks about that can money. He says if you fail, what's the big deal? You lost a few million dollars of investor money and they've got plenty more. And that's the bet they take on the chances that you will succeed. Oh, yeah. They lost a few million dollars worth of investor money. They got plenty more. Uh, Marina, go for it.

Monica: [01:15:16] Yeah, I think I'm going to kind of like reiterate some of the points, but basically so I sometimes start by saying for the startups I've been in front of and I we like to try. But the main thing that I see is that if I see a theme, they are not able to build something and get customers back, know the technology of who cares about the technology. So this started with like Data science. As a data scientist, who cares to me is if you have an idea right. Is a good idea, then you have a kind of like a good team. Right. Especially if you are at the beginning. I'm not assuming you are going [01:16:00] to have the best team yet, but at least and you can have a product or whatever like this, the MVP. So so get something that works and don't be like too white. Like I have seen so many people that try to like do like a million things. And you know for sure these people will never go anywhere. You have to be beedi like laser focus on what you want to, not trying to cover everything but laser focus. Then you get your few customers that are going to be your your your like raving fans. If AIs if you see back then, then you know that that is that has probably a lot of chances to at least move forward. Right. But the other thing is that many people come with a voice like everybody has an idea. They think that like, well, you know, I'm going to run. It's so glamorize. Right. So and then they don't even think many of their ideas are not even scalable. That is just OK.

Monica: [01:17:04] That's good. Especially I also come from academia. Believe me, there are many things that are not scalable or whatever you try to set to small thing that will never go anywhere. Right. And in terms of finances, I mean, I don't that's you know, that's something I don't know. But you probably don't want to dilute yourself too much too quick. That's that's I will say you don't you want to hold on. And the team, if you don't have a fund there that will make you so passionate, that has left everything and is doing that, then, you know, that's not going to go anywhere, no matter how beautiful it is. And how great is the idea that that's you know, it is not is you have to have a passionate person in the team that just, you know, by that will also build the culture and the mission. Right. Because that's also part of the things that I think are important is to have a good [01:18:00] mission that is going to be the culture of the company. Right. But, you know. Yeah. So to me, like the first question was about being data scientist. I mean, you can be anything. It's just I don't think is relevant. The fact that you are a data scientist or doctor or whatever, it's just that those are tools that you are going to use later. And yes, at some point, if these data center, then you need that a good tech person, a good or CTO or. Yes, but but I think that's secondary to the fact that you still need a good idea. Right. And I can focus narrow problem to solve and raising funds. Right. So your rating customers so.

Harpreet: [01:18:48] Very, very good advice. Thank you very much. My dad also loved the shirt, uh, I'm the coolest person will ever meet as a good shirt. Uh, then go for it.

Tom: [01:19:00] I saw my first business in college, Trainwreck, so I'm not speaking from a position of perfection by any stretch of the imagination in Reno in two twenty nine twenty two thousand nine, the startup scene kind of started seeping into Reno. And you couldn't you couldn't get a gold mine funded. That was the joke, is that in Reno, you couldn't get a gold mine fund. And we had everybody from the government trying to beg VCs to come in and fund companies. And it just it didn't work. Nobody wanted to fund anything that came out of Reno. So we had to kind of founders. One founder left Reno, went to the Bay Area and just founded their business where there was money and where people were looking to fund businesses the other half. And I'm one of that other half. We just stayed here and figured out how to bootstrap businesses. And our perspective on startups is [01:20:00] so different than all of the other startup beach star, Silicon Alley and all of these other places. Our perspective on startups, there's a lot of Renaults out there is not the only unique little ecosystem. We all took a completely different approach. And I mean, one of the things that we did was we let customers talk us into starting businesses. And there's a few people in Reno who have, I think the three of us, the three time founder, three time exit, and he just let his customers talk him into starting a business. And there's always need and if you are good at connecting with customers, you don't need to connect with VCs.

Tom: [01:20:42] They'll come find you because you have product, you have market share, you have passionate customers who are your evangelists that your marketing department and talking about, you know, building off some of marinas that, you know, you've got an idea that's great. But your customers really have to have the idea. Your customers have to be passionate about this thing or they won't buy it. And there has to be a business there, not just a product. You have to be able to create a business and sustain the business and do all of the things that businesses do outside of Silicon Valley, outside of the VC world, outside of the startup culture and startup mentality and start a thought process. And so remember, you're building a business, your customers are going to build it. They'll give you the ideas. Not all of us are going to start like an apple where we're telling people you need an iPhone. That's really rare. Most businesses don't do that. And everyone says, oh, those are the only ones that scale. No, not at all. You can watch a just traditional idea, organic type of business using technology of any sort scale ridiculously. If they have a passionate customer base and they build a business first and then create products that the customers love and they cultivate and grow market and do the things that a normal business [01:22:00] does. So as much as everyone wants to come at it from a startup perspective, I mean, it's just as relevant to come at it from a business perspective.

Tom: [01:22:09] I mean, pretend you're back in the nineteen thirties. How would they do it then when there were no VCs and banks told you get out, like how would you do it, how did they start businesses back then. And sometimes it's good to look at your business from a different perspective like that, because a lot of the startup stuff is hype. A lot of the startup stuff that you get pulled into in the VC stuff, you get pulled into it's noise. And if you do the fundamentals and you do the strategy and you build the business like a business, sometimes that stuff takes care of itself. And obviously, you want funding, you want VCs interested, you want ridiculous scale, you want a lot of this stuff. But at the same time, you're also building a business and it's really hard to deny somebody funding when they've got customers, got a great product and got a solid, substantial running business to say no to that person. So, I mean, always think about it from a different perspective, not just the traditional startup in v.C route. I mean, pretend you live in Reno and no one will fund your gold mine and you've got to figure it out. Sometimes that's being a founder is it's just figuring it out when when those alternatives aren't viable. Just admit they're not and move forward. Move past them.

Harpreet: [01:23:31] And very much been up and a lot of great love, great tips, great advice given today. Um, Greg, I'll toss it back to you. Let me follow up questions or comments or anything.

Tom: [01:23:43] No, I think that that that gives me a lot to to think about, um, ultimately I would like to do either or whether I continue to to help startups also found my own startup. And, you [01:24:00] know, in a situation I'm in right now, as are as I said in the comment, I think to the part local that I've been alluded to. I'm trying my best to hang on to a little bit of it. And having responsibilities doesn't make that easy. So but I do believe for anything you want to do, there's always a strategy that works where you can minimize some sort of risk, not completely eliminating them. I don't think I can eliminate them. But at some point, I think this is one of those sessions where I'm going to have to re use it to to extract some more insights from it, because I really enjoy listening to each and every one of you. So thank you.

Mikiko: [01:24:44] And I like that idea of risk, right? I mean, I think everyone has sort of a different flavor version of risk. So Robert Sudi talks about the rich life. Right. And I remember when I first read him, I was a Catholic, but the but the info was so good because I think. I mean, like my my parents were either not college educated or they were never had a college degree. Or my mom, like she's an Asian female. So I've seen all these experiences where, like, they could not there are certain kinds of risk they could not afford. One kind of risk was. Telling their bosses to go to go f themselves because, hey, you know what, my mom had like a husband to support when he was doing his consulting business and a daughter to put through college or other things like that. So, you know, for them, like for that generation, like they had a different idea of risk. For me, when I when I quit my job to go work on the startup, I had made that decision after closing sort [01:26:00] of like final or near final round interviews with like five other companies that were competitors in that space for like I as a scientist or like a normal engineer. So I'm like, OK, worst case scenario is if I fail at this, I can take that information and I can bring it to this domain space or even other companies. And I've got money and I've got a partner who didn't understand it. So, like for everyone, risk is very relative. Right. So just because someone says like, oh, this is not a good idea to me, it doesn't mean it won't be a good idea to you.

Mikiko: [01:26:32] Right. Because when I quit my job, like in the middle of a pandemic, my parents didn't like it. I didn't like the job. I want to work. My parents were like house, family, kids, like, how do you do that? But for me, it was very Inforum, because I'm like, look, worst case scenario is I go get a job or whatever. I feel that savings back up. It's all all good in the nice thing about being an adviser is you can learn from other people's mistakes and sort of like really think it through. And if it's something that you feel like very, very strongly passionate about, there's all there's tons of ways to mitigate that risk even further through the information gathering, I think. I mean, I kind of learned in some ways is that being an entrepreneur, it's like this really kind of crazy sort of roller coaster of being you have to be just optimistic enough, just optimistic and crazy enough to think the idea could work. But you also have to at the same time, kind of have a split personality where you have to be like the most negative, Nancy, in the world where you're like, I think this could put it off the rails. This could really I could get sued for a lot of money for this. So you have to be sort of like optimistic and like you have to kind of believe in your chances.

Mikiko: [01:27:50] But at the same time, really be honest with yourself about here are the areas that could ding me. How do I sort of like how do I mitigate this or how do I do that? And it's like a [01:28:00] really weird sort of it's a little bit of a weird seesaw. And I think, you know, as long as you can kind of have a foot in both of those areas, that will definitely help, as well as keeping in mind what is kind of your own personal risk curve and was your own personal journey like? I have some stuff I work on right now that people will probably look at that. I'm thinking like you're an engineer and you're wasting your time on that. Like you get paid like five or ten thousand dollars for consulting. I'm like, yeah, but I love this thing. I think about it. I dream about it. I have books about this stuff, you know. So at the end of day, I would feel so bad if I didn't pursue the idea in some form. Like it does have to be a full blown business. But, you know, if anything like if Cornton taught us anything, right, it was the fact that you don't always have tomorrow. You don't always have like a year from now. So if you have these kind of lingering questions or doubts, they can kind of like you after time, know. So that's that's kind of like my my two cents. I'm sure. It's also why it's how I'm justifying my my stuff to myself to. Right. So.

Harpreet: [01:29:05] Irrationally optimistic, right? That's the. I think, um, excellent and say excellent, excellent conversation, Greg, thanks so much for a for kicking off this discussion. Um. This is probably more than what most people learn in their MBA programs. What do you guys have been talking about today? So thank you for that great discussion. Um, any last minute questions? We start wrapping it up. Yes, remeet, that is from Sacramento. Yeah, yeah, that's true. Uh, I tried to pull the Sacramento and Punjabi and card on him to get him on the podcast. The response? It is not clear. Go for Ben.

Tom: [01:29:50] As you can tell the group, sometimes Data science becomes crazy, so Data science, we think, is going away in a notebook and doing all stuff. So I actually have to go buy a satellite transceiver. [01:30:00] All this gear in drive up, drinking a bunch of coffee into the middle of Wyoming and go hike 10 miles and fish. Not for fun, but for Data science, because there's a project coming up in three weeks, and so I'm actually going out to prove out this concept with cameras and capture and teaching A.I. to study a FLY-FISHING cast. It's just hilarious. Like, I don't know I don't know if that's useful for anyone, but just like I've never used a satellite transceiver and I'm thinking about bears. And I bet if I buy a gun on the corporate credit card, I'll get in trouble. I probably won't buy a gun. I'll get some bear spray. So it's just like I don't know I don't know if that's useful for people, but sometimes Data science could take you into wacky places that don't make any sense and not realize you are doing that for your work. It is more interesting now that but it's actually just a pretty it's a pretty high tech assessment because I'm taking a big group up there in three weeks and there's like emails with lawyers trying to figure out like risk, like actually having to lawyers about like bears and different things is really interesting.

Tom: [01:31:11] And yeah, it'll be hilarious if I don't come back, if I get eaten by a bear. Just know that there were lawyers at Data robot that had to deal with. I don't know if that it'll just be funny. So the focus is going to be on the fly fishing, yeah, the focus is on predicting if we can catch a fish on a particular cast using all of the known information outside of the casting window so that the casting, the 15 second casting window with the dry fly in addition to recency, when all of these features that people that like fishing think they know, but unfortunately. We could have done this next to like a fish hatchery, or we could have, like, floated down the Snake River or the green, but there's no hero's journey in that. So me like [01:32:00] a dumb ass. I said this story will be way more interesting if there's a hero's journey. Apparently, that hero's journey is like a 20 mile roundtrip hike. And so I'm going to get my ass kicked tomorrow and I'll return a report like Sunday. But the thing that's hilarious are the other people I'm signing up for, this ass kicking in three weeks. Anyway, some of my decisions are not well thought out, but we we see where they end up.

Tom: [01:32:22] It'll be really interesting to see what I can tell you guys in three weeks when we actually do this will either be a huge success or a bunch of like lawyers told your told you so those returns. That's right. And they would love you, but I'm pretty sure it's going to be a success. So best of luck to you. But it makes me wonder what business model can't be so right. Because when you describe this thing, you're trying to COVID the science or is there something that's complex that we call complex that science can't take a stab at? It's crazy. Like guess this discipline will work for us to do. Right? So we have and we're I'm a I'm a big fan and bringing Data science into your daily likes and passions on it. Another example would be like, if you like, wake surfing behind a master MasterChef boat, why not have Data science doing the work for you, tell you the trimm of the boat and tell you where people should sit on the boat like these are things that people think they know. They have intuition. You could actually have Data science like processing that in real time. And so my goal is how do we come up these projects, that sort of craftsmanship?

Mikiko: [01:33:27] What are you doing coming up and actually putting your knife while you're trying to create a bag or something like that, like a Bowie knife? That that's what it's about. Like, oh, put my thumb. Oh, no,

Tom: [01:33:41] I just want my droid in the future to go backpacking in the wind, reverse and fly fish for me and then fly them home all. I guess it'll be a droid, right. Or a drone or a Tesla. But yeah, it'll be the Tesla. But and I'll be in my VR space where I'm just drooling away for days.

Harpreet: [01:34:01] So [01:34:00] that's interesting. So you're going to take pretty much all of the data from the, uh, just Iot type of Data 15 second window and predict if you can catch a fish, but. You're not going to test to see fish are already there because it can be like Data leakage if you can't when when you catch the real. I'm just I'm just thinking about the actual end user application of this. Can I have a fishing reel that I tossed out that will then give me a probability that I will catch a fish

Tom: [01:34:32] Like you could? So when this is done, when the cast hits the water, you would have like a real time probability because a lot of times when you cast like you lived there for five seconds, 10 seconds, do you need to move to another side of the lake? Like, are you just wasting your time? Do you need to change a fly? So all of this stuff is up in the air while we're trying to bring this together.

Harpreet: [01:34:53] Are you predicting the probability that the fish will be there, given the condition or the probability that this particular cast will

Tom: [01:34:59] Have particular cast? So each cast is its own observation. So I've been having a lot of fun lately with these temporal windows. So in a 15 second window buffer, in addition to the non window, the non video Data, can you teach it? But this is an example, something that speaks to people. So souls. So if you enjoy fly fishing, this is like a near spiritual thing for you. And the place we're going is circling the towers. And if you talk to anyone that's been there, they're like, oh, that's an aspect. Why are you going there? I'm just like, I don't know. I don't know because things get more interesting. Right. Sounds like a lot of fun then. It's all about has to be memorable. Right. If you can check the memorable box you've won, even if there's some suffering involved.

Harpreet: [01:35:48] That sounds really interesting, refashioning your love to see how that would work when

Tom: [01:35:53] You're all invited tonight.

Harpreet: [01:35:56] I can get there, um, Sharaku being [01:36:00] in the building, who was going on my mind. Good to see here again. Um. We're right about to wrap it up, but there any last minute questions, now is time.

Tom: [01:36:16] No questions about I'd like to say good morning to everyone on the call, that's all under the

Harpreet: [01:36:23] Age of combat. Thanks for thanks for joining and sorry that we've got to wrap it up, but appreciate you dropping in.

Tom: [01:36:29] No, no, no. This does just come on Tuesday. Good morning to everyone.

Mark: [01:36:35] I actually have a quick question. Yeah, go for it. Hopefully it's quick, but essentially I'm starting to do some more teaching around Data science and various things. And I'm just curious if anyone has any books or resources not to get better at necessarily by Data science, but to get better at teaching and breaking down concepts. If you have any kind of reference points, if there's a better way to go about it, and I keep on iterating on different lesson plans, every single time I teach someone something, I take notes and like all that work, that didn't work, but it'll be cool. There's already some figured out framework, you know, that people use that can kind of iterate on top of.

Harpreet: [01:37:22] That's a good question. I don't know. But a framework where we try to look at some literature, just like, um, we just designed math curriculum and literacy, what they use with two things that I get that are coming, like there's like three reasons I would recommend that come to mind, but none of them, like, have probably that structure that you're looking for. Uh, one would be just how to solve it by uh uh. Gee, Polier Olia, um, I can get your PDF copy of that book, like, um. Have you done awesome. Yeah. Then another one I'd say maybe um. In the barbarically book [01:38:00] My Mind for numbers. But those are more books just on how to learn and how to solve problems, which you could probably take and teach those same. Things in the day, of course, they I'll flip it over to the audience and the participants, rather, if you guys have any suggestions, many or cuz

Tom: [01:38:24] I don't have any particular suggestions, but my mom is a special ed teacher so she can aspire and reach out to her. She may have some resources because she, she, she doesn't probably have. And so I can reach out and I'm sure she gets some resources. So Mark, I'll speak with you sometime next week.

Monica: [01:38:44] Mark. Oh I. Yeah, so I was an instructional designer for 11 years, and, you know, when you design courses and instructions, I think the first step is to do understand your audience, know your audience and like kind of do a little research about the people you're going to teach. I think that will make a huge difference in how you design your course, because people you need to know whether they're beginners or somewhere in the middle or if you have advanced. I think that would be the first step. I would say talk to the people that you're going to teach what they're looking for. I mean, that would be my first step. And then based on the answers that they give, how much python they know, how much machine learning they know, those kind of stuff, that can give you an idea about where they're coming from and then you design your classes based on that. So definitely the person I was talking to, the people that give you a lot of answers, Bushra.

Mark: [01:39:42] That's super helpful. Yeah, give the give further context and also saw this happening.

Monica: [01:39:48] Yeah, what happens is most people jump into the content before even reaching out to the audience. I mean I mean, I've taken a lot of data science courses. Some of the some are just so advanced and some are like in between, [01:40:00] partly because this is what I'm talking about. So they have the Udacity and they have this way of doing teaching classes. Right. And you take it in the audience and it's so big and fast that you have beginners and advanced and the find it's too hard, the advance is too easy and people in between finding a way through. So yeah. So coming from an instructional design perspective, talking to your audience and seeing what they want, it's crucial. So that'll really help in formulating your courses and you'll be super successful knowing what they want. So.

Mark: [01:40:41] That's super helpful. I think you hit on the point out I'm currently struggling with is that I'm currently mentoring four people and that's like my first practice. I like doing the teaching and whatnot. A couple of them are like really beginner new. And a couple of them are like really events with sport between. It's been very challenging. It's been some lessons learned on my own. So this is so helpful.

Monica: [01:41:06] Yeah, I used to work for a digital entertainment company and I supposed to train. We have all this digital entertainment products that come out of the company and and we had to kind of design courses based on technical fields and then based on the end users and stuff like that. So end users, they don't want to know the technical jargon. So you got to really kind of change that will change the whole technique, looking into more end user type of lessons. So any technical then, of course, it will be more technical, but still has to make sense to them. So I learned the hard way of not just designing versus not knowing my audience. Sometimes people don't like the process of previews and so sometimes people love it, but knowing it's supercritical.

Tom: [01:41:53] Yeah, maybe if I may I talk about all this, but generally what I found to be [01:42:00] quite useful is maybe one or two books on classroom management. If you are doing classroom style kind of teaching, that's that's one. The other one is actually it coincides with what I'm currently still researching on, and that is how people learn cognitive science. So I think it would be good for you to pick up a few. I, I can remember there was this there was this book with the title How we Learn Something along that line. Probably I need a bit more time to see if you can maybe you can drop me a message on LinkedIn or something. Once I found the book, I can I can show you that book is pretty good in the sense that it actually shows like how we humans actually learn stuff and it's something you can pick up and even maybe help them with certain learning techniques. That is even useful after the class as well. Uh. So to put it in a nutshell is that we humans, we learn differently, and that's a cycle, that's the cycle of picking up and as a cycle of sort of letting the multiverse all settle in your mind. So that's that's that's the two phases of learning iodines. A pretty interesting message, I think. And then I can share more with you.

Monica: [01:43:29] Sorry. The other thing it could do, you could design a small force and then just let people test it like you get a few people and let them look at your horse and then they will give you feedback what they like, what they did not like. And I did that quite a bit, especially when I was designing, training, learning websites, as you can have a bunch of people. And these people can come from different groups. It could be the executive team that could be the actual people with the training materials and then the managers, they all will give you different things [01:44:00] to different things. But finally, at the end of the end result will be who is the support for? Those are the people that you want to target and hone in and make a good, I guess.

Mikiko: [01:44:12] And Mark, you said that these were these are like and kind of like Newby's a little bit

Mark: [01:44:18] For my for my mentees newbies, but for this there's more formalized course. It's relatively new people to in analytic space. OK.

Mikiko: [01:44:28] I think something that that really helps just also some of the resources I've seen just go wild, like, for example, made with Mel. That's like one really, really popular resource, like in the space. There's a few others will be boring. All that some of the guiding principles, I think, that are really. So our first off, build intuition before you go into jargon.

[01:44:55] So there's this.

Mikiko: [01:44:59] Rightly so, I think a lot of engineers or technical people, they can sometimes get too much into the details and a lot of times I think we're probably new people struggle. And we're I certainly struggled was being able to place the information into like a concept map or into like a development workflow or things like that. So I would say stop from start from high to low and then frame all your sort of like try to frame your information as much as possible around why. So Eric posted about it in comments, talked about how he kind of comes back to why. And I think the reality is that for me, like when I'm learning, if he if I have no value, if there's no reason for me to know information, I will not remember it. So if you can kind of essentially guide them along the journey of in a way like they're asking themselves why and you're content in your course is providing the answer, then I think that [01:46:00] will also go a long way towards helping because I see some courses where they just literally copy paste from like the Tablo website. And I'm like, I mean, why the hell should they pay for your material? Right. Like they are coming to you for a reason. And that reason is probably because they need a further sort of like breakdown as to like the content or the problem or the topics at your courses targeting.

Mikiko: [01:46:23] So I would say like start high, then go low building tuition before you go to code examples and try to connect it back to your experience, like working in an industry. Right, because. And I think everyone's had this experience right, where someone puts up a hypothetical example, right, like like Sally has asked Barry for a thousand dollars to do X, Y, Z, and then you split and you do compound interest and blah, blah, blah, blah, blah. Well, OK, that's great. So once again, they could probably copy paste that kind of textbook example. But where you could kind of added value is you could you don't have to do the full details of stuff that you experience right at work. Right. And all that. Right. But what you could do is you could sort of frame it in like, you know, here is like write the star framework or whatever. Right here is a situation. Here was what the problem was here. The relevant facts did it right. Like but we've those sort of concepts in two books, I think that do a really good job of that. Let me talk to my library are actually like a business case study books, because they're really good at those kinds of like building those stories such that you have a problem to solve.

Mikiko: [01:47:38] So one of them is point at point. Yeah. And then the other one is a case study handbook. From elite or it's one of the Harp is a school, and I think those two are really good examples of stories and how they sort of like frame stuff up. [01:48:00] So I think, yeah, I think those would be like three pieces of advice. You definitely don't want to do something like generic that people have seen a million times. And like, the best way to make it not generic is to incorporate your own sort of experiences. And your wise is especially around why that information is important. Right. Why is it important to know as a test? Why is it important to understand the distribution? Why is it important to make sure that your data is clean and you can pull in? Real examples of your data is not clean. If you have bias in the data set, well, that's how you get banks that lend out loans on different interest rates, depending on the color of the receiver. Right. That's something that people can relate to, like, oh shit, this is really not good. So.

Mark: [01:48:44] This is also helpful. Everyone, this is this is really great. Yeah, just try new things and learn as I go along.

Harpreet: [01:48:53] And let's go ahead and wrap it up. Great questions today. Great session. Thank you for joining in. Thank you for sticking around almost two hours and shout out to everybody that's been watching on LinkedIn and everything. Don't forget to tune into the episode are released today with David Benjamin. It's all about cracking complexity. His book is It's a great book. And if you have the audible premium membership, whatever it is, it's available for free on that. So definitely go get that book. And then also did an episode earlier this week with all tensioner that's going to be released next week. So I definitely go ahead and get into that. Guys, take care of the rest of the evening. Remember, my friends, you've got one life on this planet. Why not try to be something good, everyone?