OH28-29-08-2021

Harpreet: [00:00:06] What's up, everybody, welcome, welcome to the comet Amelle. Open office hours, it is Sunday, August 29th. Three more months left in the year. That's crazy, man. Just thinking about that, how fast this this year has gone. Four more months. Yeah, I guess. Yeah, I guess it's no October. November, December. Yeah, yeah. Four more months left in the year.

Harpreet: [00:00:25] I hope you guys get a chance to tune in to the podcast episode that I released earlier this weekend on Friday with Jeff Li. Um, that was a cool episode.

Harpreet: [00:00:34] He's a real cool guy.

Harpreet: [00:00:35] He spit a freestyle flow for us. So if you cut the first 30 seconds of the Sizzler, you can hear him busting some rhymes for us. And then also, I was actually on the Kanji podcast, or rather Ken's nearest neighbor podcast earlier this week as well. He released that on Wednesday. So that's a big

Harpreet: [00:00:51] Conversation hanging

Harpreet: [00:00:53] Out with Ken. So it's always a good time. I hope you guys get a chance to check those out and stay tuned for another couple of awesome episodes

Harpreet: [00:01:01] Coming your way over the next

Harpreet: [00:01:02] Couple of weeks ago. Want releasing with Max Frenzel. He's an AI researcher and we talk a lot about burnout. When I recorded that episode, Max, it

Harpreet: [00:01:12] Was I think it might have been

Harpreet: [00:01:14] February of this year when we recorded and that time did. I was going through some really, really severe burnout issues. And so it was nice to just kind of open up to him and talk about that and get some tips on how to deal with that. But if you guys are watching here live on YouTube or on LinkedIn or whatever it is that you are joining us from. First of all, you are more than welcome to join us right here in the Zom room and be part of the conversation. All you got to do is click on

Harpreet: [00:01:38] The link that is in

Harpreet: [00:01:39] The description or the video. You can join us for those who listening to the podcast, wondering how you could join us in the future. Simply go to HTP colon slash, slash it dot l y forward slash commet dash lmld dash 08 and you can register for future afzar sessions. So if you got questions wherever you're watching, go ahead and drop [00:02:00] them right there in the comments

Harpreet: [00:02:01] Section or right

Harpreet: [00:02:02] In the chat if you're if you're here and we'll get to them. So let's go ahead and start taking some questions. Christof, I know

Harpreet: [00:02:08] You said you had

Harpreet: [00:02:09] A question cued up for us here, man. So let's dig into it.

Harpreet: [00:02:12] Everybody else listening.

Harpreet: [00:02:14] All questions are welcome. I see a few new names coming in here, if you would start. Prashanth, what's up? Depak, what's up?

Harpreet: [00:02:20] All questions

Harpreet: [00:02:21] Are welcome. And any questions you guys got, whether it's three related or activity related, life related, whatever, man, everything. It's all it's all fair game here. Cristoff, go for it. I'm not able to hear you, unfortunately.

Harpreet: [00:02:34] So there's some some audio issues, if not.

Harpreet: [00:02:37] In the meantime, Chalco Kalka. But I know you had a

Harpreet: [00:02:39] Question that you're that you're burning to ask me

Harpreet: [00:02:43] Or something that you want to discuss. So by all

Harpreet: [00:02:45] Means, man, how could we go for it if you're still still there?

Harpreet: [00:02:49] All right. Well, it looks like Kalka is not there and Cristoff is not there. So is this that

Kawkab: [00:02:55] We're down here.

Harpreet: [00:02:57] All right. Good night and good to have you back, man. What's going on?

Kawkab: [00:03:02] I hear the double here as well. I turn the wheel of what I said. I spent three months building up my machine learning engineer. And that's a good.

Harpreet: [00:03:17] Hey, man. Congrats, man. That's awesome. I know you've been working really, really hard to learn. Sarfati, I'm very happy for you. Where is this that we're going to be working at?

Kawkab: [00:03:26] Well, for now, it is very remote, but it will be a very important mission. Show the information that came in that gave me the offer letter and it said that I would be working the city back up to make this project that is

Harpreet: [00:03:48] Ok for like a contract kind of contract to have a

Kawkab: [00:03:52] Contract job. But they said that the contract is the minimum of three years and and [00:04:00] it is paying me. I told you about the pay that they gave, including benefits, benefits and bonuses. All of that is there.

Harpreet: [00:04:10] That's awesome. And granulation. And I've been working really, really hard. And you're going through tons of applications and interviews, just persistence and just getting better and better at the process. Yeah. Just crushed it, man. So very excited for you. That's awesome.

Kawkab: [00:04:23] You've got the hybrid. They gave me the occulted and now they they're saying they need minimum of to three weeks to onboard me. I think this is my first job. But is that is that usual or something is going on there?

Harpreet: [00:04:45] No, I think that's pretty.

Harpreet: [00:04:46] Pretty typical. Yeah, I

Harpreet: [00:04:48] Just I logged into my email address. I got a bunch of meetings for for onboarding and stuff like that happening. And I think especially when it's everything's remote now, it makes makes more sense to extend that process because you got to deal with multiple calendars and stuff. But I think.

Harpreet: [00:05:01] Yeah, onboarding, you know,

Harpreet: [00:05:03] Onboarding, really. Again, you were up to like four to six weeks because you have to learn

Harpreet: [00:05:07] The ropes of everything. Oh, you know what?

Harpreet: [00:05:09] You got to learn the team. I learned to project what the data's out, what going to be working on things like that. So that's a typical development. Unless there's unless they're trying to say unless you're trying to say that they're not going to pay you during the onboarding

Harpreet: [00:05:20] Period, then I'd be

Harpreet: [00:05:22] That'd be a bit suspicious about that as long as they're paying you.

Kawkab: [00:05:25] They don't talk about that. They did not talk about whether they will pay me or they will not pay me.

Harpreet: [00:05:32] They will.

Harpreet: [00:05:33] They typically will. The onboarding process is just, you know, just you getting familiar with the with the ins and outs of the company and the team and and things like that.

Kawkab: [00:05:42] So and then they to the operator. And if you are going to hire a leg, I would then go on Wednesday and they give me a call back on Friday. They said that they want to meet me for 20 to 30 minutes and [00:06:00] instrumented room. And I met them and they give me a call at 5:00 in the afternoon. And then the recruiter called me there and make you understand your concern. But they are going to do the background check and the drug test. So that process usually takes three weeks.

Harpreet: [00:06:22] Yeah, that's pretty pretty standard stuff. And I mean, it they probably really liked your portfolio projects. And I've you know, I've seen your projects that I've been working with you for quite some time. So there's no doubt to me that that they wanted to move so quick to lock it in. So that's that's I mean, that's atypical to get an offer that quickly. Usually it might take, you know, a week or two distinct means that they're really,

Harpreet: [00:06:48] Really

Harpreet: [00:06:49] Bullish on you, meaning they really were, you know, have have high hopes for this. I think that's great. And, you know, just don't do drugs and you'll be fine.

Kawkab: [00:07:01] But I smoke in my life, so I don't have that at home.

Harpreet: [00:07:06] Right. Well, congratulations. We got a congrats coming here in LinkedIn. Dustin, you're saying congrats, Kout Covid. That's awesome. So, you know, that's you know, I'm happy for you and everybody else listening intune and happy for you, too. I mean, me especially, man, because we've been working together for a long time. So I know

Harpreet: [00:07:23] How much effort

Harpreet: [00:07:24] Now you've been putting in. So congratulations, man, the big time for you. I don't feel free to hang around. And if you have questions or if you have comments or anything, please do let us know. Cockup. Let's move on to Crystal's question. Let's leave. Krishan Cristoff has any audio issues. If if not, I think I said issues wrong. If not, then we got some questions coming in on LinkedIn as well.

Harpreet: [00:07:49] Cristoff, go for it. Nope.

Harpreet: [00:07:51] This is a

Harpreet: [00:07:52] Strange. I know what's going on, man.

Harpreet: [00:07:54] What have you have you tried turning it off and trying to get back on? I don't know. Let's [00:08:00] see if I had asked you what happens. If not give it a shot. Why don't you go ahead, toggle this stuff and just try to see if you can jump in and we'll get to you. But we'll move on to some other questions coming in from LinkedIn here. How do you this come from Christian Christian? What's going on, man? He's also I've got a podcast of his own good friend, Christian. How do you suggest that navigating balancing deadlines and Data

Harpreet: [00:08:25] Debt creation

Harpreet: [00:08:26] With the business is the Data engineering question? Let's see, how do you how do you suggest navigating balancing deadlines and Data debt creation? So what do you mean by Data debt creation, like technical debt, especially if you could jump into the room and end

Harpreet: [00:08:43] Up talking

Harpreet: [00:08:43] About this a bit further? Manabe super, super helpful. I'll actually go ahead and share a link to the to the resume call right there on LinkedIn. If you could join in, that

Harpreet: [00:08:52] Would be awesome.

Harpreet: [00:08:53] But I'm assuming you're saying technical debt is what you mean. So if that's the case, I mean, navigating, navigating, balancing deadlines. All right. So deadlines are always kind of hard to meet. And I think that's just all about prioritization.

Harpreet: [00:09:08] Right.

Harpreet: [00:09:09] So there's that matrix, what's called the Eisenhower Matrix for prioritization. That could be helpful. But for me, it's always been it's what is the most highest impact thing that I could be working on at this moment that is going to push the needle forward for whatever thing the company is trying to do and to focus on that. Right. So if you've got multiple stakeholders coming to you with multiple requests, you're going to have to talk to them and figure out what is

Harpreet: [00:09:37] It that is the most impactful. Right.

Harpreet: [00:09:40] Like if you're going to work hard on something, you're going to push

Harpreet: [00:09:43] Push through on something and

Harpreet: [00:09:45] Make sure that whatever you're doing is, you know, to have the maximum impact and Christianizing that you're in the room so we can we can go to you so you can clarify the question to make sure that, hey,

Christian: [00:09:55] Awesome. Can you hear me?

Harpreet: [00:09:56] Yeah, I can hear you perfectly fine. Awesome. So some people in the chat off. [00:10:00] I'm joking.

Christian: [00:10:01] Yeah. No, what I what I mean by my question. Harpreet, appreciate you inviting me on to, by the way. So hi, everyone. I'm Christin Steiner. Data engineering is kind of my realm right now, my world. So balancing not actually having a proper Data model, whether it's star schema or Kimbal methodology from a data warehousing standpoint, we're just dealing with raw transactional data right now. And my current world, I'll keep it high level. But having that proper dimensional model in place would really, really help us. But when the business is pressuring you to hit all these deadlines, you know, we end up with these nasty, you know, multiple subquery, you know, enjoins left joints and we just roll that into our bi

Harpreet: [00:10:49] Tool and

Christian: [00:10:50] Visualize it. But that's not really how it's supposed to go from. The tool that I utilize is looker. So really, you're supposed to have a proper model, Data model in place and utilizing it. So we're pressed for deadlines. So I find myself in this constant battle of trying to explain to the business that we need more time or that we need to develop a Data model. But they also have the priorities coming downstream from the executives. So I guess that's that's kind of where I'm going with it. And I just suggesting how to how to navigate that or communicate it in a way that the message kind of gets across smoother, maybe. But yeah, that's that's that's kind of what I had in mind.

Harpreet: [00:11:28] So are there any reports that you have in the organization

Harpreet: [00:11:32] That are created

Harpreet: [00:11:34] Kind of on an ongoing, consistent kind of periodic basis? Yeah. If you do have those reports like that, can you calculate the time that it takes for manual effort to create one of those reports? Right. And then you could say, look, right now we've got this one report that you guys are looking at every week, but it takes this you know, it takes three of us five hours a week to collectively put an effort just to create this report. Right. And if we're doing [00:12:00] this every week

Harpreet: [00:12:01] Because it's manual, you know, that's

Harpreet: [00:12:03] 15 labor hours over the course of 52 weeks, because, again, this report 15 times it cost this amount of money. Right. You know, you could say that it cost 15 times, 52

Harpreet: [00:12:14] Times, whatever average

Harpreet: [00:12:15] Hourly rate. And that's just the dollar value fixed that. And you can say, all right, this is what it's costing us to create these reports in terms of time. Now, if we wanted to try to put in some effort upfront right now, maybe we spend 40 to 60 hours creating a pipeline that takes this transactional data and just automate the aggregation and summarization of it puts into a data warehouse that it just plugs in nicely into what it looked at in your case. Then we put in that effort upfront. Now, all of a sudden, you get this not only every week, you're going to get every hour, every minute, whatever freshness you want, the Data, you can have it. And we don't have to do this over and over every week. Right. So you can just automatically come in with a kind of a reduction in cost just in terms of time for putting in that effort. I think that's probably right. No, I multiply that multiply that by however many reports you guys got that you're working on. So that's how I would approach it. I mean, it's a very similar situation

Harpreet: [00:13:13] That some every

Harpreet: [00:13:15] Company has its top issue, right. Everybody wants everything

Harpreet: [00:13:18] Yesterday, but they don't want to spend the

Harpreet: [00:13:20] Time to get it in place. And they're just like, what is all this stuff? What we have to do, all this blah, blah.

Harpreet: [00:13:25] But you want to

Harpreet: [00:13:25] Make it look and feel like nothing changes for them. Just explain that to them, that you don't have to do anything different on your end. It's just us on the back end, streamlining stuff, making stuff cleaner and neater, easier to troubleshoot, easier for us to add in any visuals or any summaries that you are interested in. We could do that real quickly for you now instead of having to redo everything, rebuild the wheel type of thing. So that's how it approach that.

Christian: [00:13:51] I love it and I love actually quantifying the bottom line dollar that it's costing them now. So. Yes, I'm dealing with a quarterly report right now. So [00:14:00] that's as far as I'll get into it. But yeah, that's it's it's just about a challenge that we found a hiccup on Friday. And it was it's like the deadline is Friday. It was Friday. And so now all of us engineers are like scrambling to have to include this in join potentially. You know, it's it's like a two subquery inner join it because it's just wild, because we're utilizing the pull. This is transactional. So it's very dynamic fields that can change depending on how you subquery itself. Yeah. It's just that it's it's been a lot of learning. A lot of learning, I think. Sure.

Harpreet: [00:14:35] It sounds very, very familiar to me. Let's go to let's go to go to Austin. Austin Gopher. And then there's some great comments coming in from Dustin that I'll read off coming in from LinkedIn here, Budgeter Austin.

Austin: [00:14:48] Sure. I think there's like a more general sort of premise here, which is I think no matter if you're working in the kind of role you're working in Christian or what I do sort of in this like kind of community and trying to this communication to business stakeholders is like to find ways to. Like to speak in their language because like the thing, the struggle is like these kinds of things are deeply personal, like you're having to experience this thrash and this difficulty in getting your project done. So to you, it feels like I can't like this is personal, like I can't do this work, my team can't do this work. And so that

Harpreet: [00:15:20] That bit like

Austin: [00:15:21] Figuring out a way to turn it around and pitch it to the leadership or whoever it is in a way that speaks their language. So it's like bottom line dollars or excess resources they're spending or,

Harpreet: [00:15:34] You know, showing them what

Austin: [00:15:35] The before and after it looks like somehow, some way that puts it in their context as like business leaders and is empathetic to like what they have to do. They have to report to and who they have to sort of like work it up the chain. It's not always easy, but I think like the more it can become depersonalized and about like the goals of the organization writ large, it's like we're doing this thing at comment or we're sort of looking at our documentation and [00:16:00] thinking about a refactoring of it. And, you know, instead of just like having going off a hunch of how we feel, the way we're getting green light on this is to go, you know, do a survey with our users and do some iida on that and do this whole process so that when we go to the leadership, it's like we're speaking their language about like, you know, we're responding to this user feedback and we're going to execute this in this very particular way. So having that sort of ammunition that speaks in their language, I think is super important, no matter what you're doing, whether it's a Data position or something else. But I think in this case, like I think Harp Reid's exactly right, is putting together like a bottom line calculation is speaking their language. So that's the general point, is that like what is it that they need to hear to to implement that into their view of the company or in their contexts, I think is a general point is important.

Christian: [00:16:47] Know I appreciate that so much to being newer, to Data engineering. You get so in the weeds with, you know, because I'm like trying to learn about data warehousing right now, you know, learning about the yellow behind it and then looker on very admin level, something I never had to worry about as an analyst. I get so in the weeds. I forget to kind of take a step back from that business context, which seems obvious, but I guess it's it's not all the time for me. So appreciate it.

Harpreet: [00:17:12] And I'll read out some comments coming in from LinkedIn from Dastan. Then we'll go to our Paul. So best in saying I try to spend some time each Sunday evaluating the best way to spend a majority of my time for the week based upon what will add most value to the business. I really like that approach. That's kind of similar to the approach that

Harpreet: [00:17:31] That I do

Harpreet: [00:17:32] With these with these that have like a blank sheet that I that I fill

Harpreet: [00:17:37] Out every

Harpreet: [00:17:38] Sunday and just talk about what it is that I plan on accomplishing over the course of the week. I get to fill mine out. Dustin also says dedicating time to things that will have a high long term return, such as automation,

Harpreet: [00:17:53] Fall into the important

Harpreet: [00:17:55] But not urgent category,

Harpreet: [00:17:58] Which it's so

Harpreet: [00:17:59] Important [00:18:00] to block and dedicate time towards regularly. I think he doesn't very much and doesn't buy by all means. Pablic that link right up there that join us in the room if you're free. Poor man. Good to meet you. I don't think we've ever met. Thanks for coming in and hanging out with us. Go for it.

Paul: [00:18:15] Yes. Thank you for having me. This is really cool. I just logged on to LinkedIn and I saw people talking about data sciences. Like I was like a Harp in for a sec.

Harpreet: [00:18:25] Awesome. Yeah. Happy to have you.

Paul: [00:18:26] Thank you. Yeah. So I was curious about transitioning from analytics into data science as a path rather than just going for a masters. I didn't plan

Harpreet: [00:18:39] On doing

Paul: [00:18:40] A Masters, and I thought it would be like smoother to go from a data analyst into a data science position, because then I would have been working with Data for a while. But do you think that step is even necessary or if so, I'm a student right now. So and I graduate in a year and I'm studying data science. So when I graduated, I was planning on just going for the analyst job first and then trying to go into data science. But do you think that step is necessary or should I just apply straight for data science positions?

Harpreet: [00:19:14] Yeah, you can kind of point in that. I think definitely start applying for data science positions. Why not? But then also apply for it for Data analyst positions. The question about do you actually need a master's to get into data science? I don't necessarily think so. It will be helpful, absolutely.

Harpreet: [00:19:30] But like those kind of dig a little bit

Harpreet: [00:19:32] More into that. So you're currently a student. Are you also working as well, or do you have prior work experience or are you a student in graduate school? Student in undergraduate school?

Paul: [00:19:41] I'm an undergraduate

Harpreet: [00:19:43] And I'm currently

Paul: [00:19:45] Working for a Bluebonnet Data, which is like it does. We do data analytics for political campaigns. And I also think I'm going to start on a research project.

Harpreet: [00:19:59] I just [00:20:00] in an interview last week

Paul: [00:20:02] For this

Harpreet: [00:20:02] Semester.

Paul: [00:20:04] So that's a possibility.

Harpreet: [00:20:05] And they're both

Paul: [00:20:06] Kind of analytic space, like using Tableau. Yeah.

Harpreet: [00:20:11] Is that an episode on my part,

Harpreet: [00:20:13] Because I think you'll really enjoy was with Stan Lewis and Sam Lewis does. Uh Data. Science on political type of Data. So we talk a little bit about that, like how did he feature engineering and stuff like that?

Harpreet: [00:20:26] So it's definitely an episode

Harpreet: [00:20:28] Worth checking out. I think in your case, Manlike, you're in Verdie working with Data and you're about to graduate. I would start just applying for Data science positions. Right, simultaneously apply for Data analyst positions, see whichever ones you get callback back from, you know, see how you progress along to the interview process.

Harpreet: [00:20:46] You know, don't

Harpreet: [00:20:46] Not apply for the data science positions. Just you should apply for those. But like, if you got the fundamental skills right, there's obviously the quantitative background that they need. But then also just the technical skills of like, you know, SQL, you know, Python or.

Harpreet: [00:21:02] Right. And you go from

Harpreet: [00:21:04] Data to a decision

Harpreet: [00:21:06] Without getting stumped. Right. Like, you

Harpreet: [00:21:08] Know, like a you have a principled workflow or executing on a Data project.

Harpreet: [00:21:14] Um, having those

Harpreet: [00:21:15] Things in place, I think are much more important than than the actual degree that you leave school with. I'll pause there to look to to let you ask any further

Harpreet: [00:21:24] Questions or let me know if that was

Harpreet: [00:21:26] Helpful or not.

Paul: [00:21:27] Yeah. Yeah, I know that that's helpful. So in terms of applying, do you suggest. Because I do have this thing with my school where I can apply, like recruiters reach out to the school specifically and then like there's LinkedIn, what's the best way to apply? And also, I'm graduating in a year or so. Should I

Harpreet: [00:21:49] Apply now or

Paul: [00:21:50] Should I wait till I get closer or relocated? And then Berkeley, I go to UC Berkeley.

Harpreet: [00:21:57] Oh, nice, nice. I'm actually from Sacramento, so right down [00:22:00] the street ish. Kind of right down A.D.s where I'm from.

Harpreet: [00:22:04] Yeah. Area.

Harpreet: [00:22:05] Yeah. Yeah. So I think in Bay Area, like you got an abundance of opportunities out there. And I think there might even be an abundance of opportunities for internship peper roles as well. So I think maybe from now until the next like four to six months, I'd apply for internship roles. And then once you're like six months from graduation, then start applying for those full time roles. I think that's a good kind of cadence to to do it. I know Kalka was here talking about how he got hired in like a day. That's not typical typically. Typically, the job search process can take a few weeks,

Harpreet: [00:22:41] You know, like six to

Harpreet: [00:22:43] Eight weeks for for interviewing with the company. Like from the time you apply to the time actually getting an offer, it could go up to like eight weeks realistically. Okay. So that that is a is

Harpreet: [00:22:54] A time frame.

Harpreet: [00:22:56] Chalco wants to chime in here. So calculable. Go for it

Kawkab: [00:22:59] For for for a while. I said, Stenseth, you are still in school, right? For me, I mean, you don't have to totally ignored my advice or recommendation. All right. Now, Baracoa, the job market is a poor example. I was at Half-Breed for a very long, long year. I made corporate life in in the beginning of January. He was pushing me to keep applying dpf man. Well, that is true. But my experience is that right now nobody wants to hire a muscularly Intiman is done. Well, that is my experience again. So in this book, I do get some kind of White-Collar developer internship where you would give them a hand. So I would activity like you are building more than you are deploying into AWB [00:24:00] is deploying it, you know, and once they put those on there their summer, you will see somebody's recruiter will protect you, even if you're being

Harpreet: [00:24:11] Something you want to

Kawkab: [00:24:12] Be. It may be there, but they will swear it never will go to court. And we need someone to rescue me. We don't care if you have ever experienced in massive letters since they were kids, you don't even need one. If it is to really put that on that project, put that on the resume, and then you'll get tons of recruiters, oil content. You don't have to look for it. And if it happens, will be over there for weeks to go. I did that up. Neighbor added jobs, but to talk with recruiters, the life lessons in law with so

Harpreet: [00:24:59] So that that brings me brings me to my second point. I thought he was touching on another point that what I mentioned is projects that you have to do some projects as well. Right. So as much as you can. In the next, you know, three hours now until a year, you know, you want to

Harpreet: [00:25:14] Make sure you do a bunch

Harpreet: [00:25:16] Of projects,

Harpreet: [00:25:18] So make sure that the

Paul: [00:25:20] Projects with models like because I have some data analytics projects like using Tableau, but it's not really applying any machine learning yet. Yeah. Those are how to get ready for data science rules. Use of the models.

Harpreet: [00:25:36] Yeah. So I would

Harpreet: [00:25:37] Recommend like three

Harpreet: [00:25:39] Types of projects for anyone's portfolio. Like, I think these three types of projects will cover the depth of experience that you would be required

Harpreet: [00:25:48] To have or the

Harpreet: [00:25:49] Depth of skills you'd be required to have as as, you know, somebody breaking into data science. First is just a project where maybe all you're doing

Harpreet: [00:25:57] Is like the Data engineering

Harpreet: [00:25:59] Type of project, [00:26:00] like something where you're pulling data from somewhere, some source. It could be just getting daily data from the Weather Channel. Right. Api. And then we want some manipulations and aggregations or whatever you want to do. You know, combining that with other data sources and then pushing the aggregated tables to either a local database or maybe a cloud database. Right. Data engineering type of project. We can eat essentially an itol project. Right. And second type of project, I would suggest, is one that's heavy on exploratory data analysis, heavy on, you know, statistics and

Harpreet: [00:26:36] Things like that. Right. So that would be kind

Harpreet: [00:26:39] Of more of the analytics ish type of

Harpreet: [00:26:42] Project.

Harpreet: [00:26:43] So heavy on exploratory data analysis, statistics, things like that. And then the other one would be just an end to end like machine learning project. Right. And that's just,

Harpreet: [00:26:53] You know, building out,

Harpreet: [00:26:54] Building out a model and everything from coming up with the

Harpreet: [00:26:58] Problem, you know, problem

Harpreet: [00:27:00] Framing design. Good question. Finding the appropriate Data. And then once you get the appropriate

Harpreet: [00:27:05] Data, doing whatever Data

Harpreet: [00:27:07] Cleaning steps you need to do, then building out the model well, doing the exploration, exploring your Data, building out the model. And then

Harpreet: [00:27:15] If you want to put the icing

Harpreet: [00:27:17] On the cake

Harpreet: [00:27:17] For that, I deploy

Harpreet: [00:27:19] It. Either you make it so that it can be deployed either locally just in your Web browser on local host or I deployed in in on a W AIs or something like that, like how cop was mentioning that way here. And if you do that with this much real world data as possible that way, you're showcasing not only that,

Harpreet: [00:27:38] You've got the technical

Harpreet: [00:27:39] Skills in terms of education, but you could actually do the work. Ok.

Paul: [00:27:44] If it's OK, add a follow up question. Yeah. Sure. So I have done a project that's kind of similar to the last one you mentioned. So I did a K nearest neighbors project where essentially I watched a [00:28:00] dating show on YouTube and I typed in all the pickup lines that people were using.

Harpreet: [00:28:06] And then I

Paul: [00:28:07] Created like an algorithm to predict the f the efficacy of the pickup lines.

Harpreet: [00:28:14] And it works

Paul: [00:28:15] With the sixty three percent accuracy. And so, first of all, one question was, I didn't use it. I kind of used it using the tools that I learned in one of my classes, which aren't like industry standard tools. They're just like built in from the Berkeley API. And then so that has to do with like, yeah, that's kind of what I wrote in the summary. Like at least it's better than just guessing. But that is my second question, which was

Harpreet: [00:28:43] How do I like know how good

Paul: [00:28:45] That algorithm even is? Because I just I don't know what to compare it to. Like, how do I address like how effective that even is?

Harpreet: [00:28:52] Yeah. Feed me a

Harpreet: [00:28:53] Baseline.

Harpreet: [00:28:54] You need like

Harpreet: [00:28:55] The you need to have a baseline algorithm or baseline measure of some sort of baseline prediction. And then anything you do that's more complex than that baseline. Then you can proclaim that you have made an improvement or that machine learning is a suitable task for this type of problem. So, for example, like, you know, let's just say you to do in your example is tough to come up with the baseline without have to think about that and work through it. But let's just say you do like a linear regression.

Harpreet: [00:29:22] Or let me rephrase.

Harpreet: [00:29:24] Let's say you're doing a regression problem.

Harpreet: [00:29:26] Right.

Harpreet: [00:29:27] But we're doing a regression problem. Maybe a baseline model that you can build is just a simple linear regression.

Harpreet: [00:29:33] Right.

Harpreet: [00:29:33] Simple linear regression with

Harpreet: [00:29:35] All of the variables.

Harpreet: [00:29:36] Input. Right. And find out. Or even simpler than that. Maybe even simpler than linear regression. You could say, you know what, I'm just going to predict the average value from the training set on all unsign data. Right. And make that my baseline model.

Harpreet: [00:29:53] Right.

Harpreet: [00:29:53] Simple baseline that that would give you an answer.

Harpreet: [00:29:56] And then, OK, once you do that, you're going to

Harpreet: [00:29:59] Have your mean [00:30:00] squared error or maybe whatever it is that you want. And then from there, build out more and more complex models to see what does

Harpreet: [00:30:07] Better than that baseline.

Harpreet: [00:30:09] They always need to have a baseline. So I recommend checking out

Harpreet: [00:30:13] A drop

Harpreet: [00:30:14] A link in the chat, but there's Jason Brown. Lee from Machine Learning Mastery has a great

Harpreet: [00:30:20] Article about just the

Harpreet: [00:30:21] Importance of having baseline models in machine learning. So I highly recommend checking that out. But then also math, just like that question that you're solving, like that project, that problem statement like that's super interesting in itself. That would be enough for people like, oh, that's pretty interesting. Let me bring this guy in for an interview to talk to him about that project. Right. Well, this is the importance of having really unique, different, interesting projects that you just find interesting, right? Right. And as long as you're able to paint a story about why he did that

Harpreet: [00:30:50] Project and and

Harpreet: [00:30:52] Talk about it in an interview, make it interesting, and then walk somebody

Harpreet: [00:30:55] Through your entire

Harpreet: [00:30:57] Process. How did you collect data? How did you clean the data? You know, why can't your neighbors and your entire file process talk about that not only in the project in like an executive summary, but if you talk about it in the interview itself. That's Goldman. So, yeah,

Harpreet: [00:31:14] I like that.

Paul: [00:31:15] Thank you for all this great advice. Is this something that you do normally on Sundays?

Harpreet: [00:31:20] Yeah. So I got the I got two sessions. So this one is every Sunday. I'll drop a link to registered in the Chadash. Yeah, there might be a

Harpreet: [00:31:28] Link to register in

Harpreet: [00:31:30] In the comments, but I'll drop a link to register there. We got one on Sunday and I got one on Fridays

Harpreet: [00:31:35] As well for

Harpreet: [00:31:36] Thirty p.m. Central

Harpreet: [00:31:37] Time.

Paul: [00:31:38] That's that's awesome. Thank you so much.

Harpreet: [00:31:41] Yeah, definitely. And you know what? It gets messy experimenting with machine learning. I suggest you check out commet IMLS experimentation management platform as well, because I think the sooner you get comfortable with managing your experience experiments, the easier it will become when you're in the industry and [00:32:00] working. And it's just part of a more principled workflow as well. So definitely keep an eye out for more

Harpreet: [00:32:06] Awesome content

Harpreet: [00:32:08] Just in the

Harpreet: [00:32:09] Near future. That sounds good. Thank you.

Harpreet: [00:32:11] Right on who feel free to hang around by all means and check in the chat here. Natasha, how's it going? Natasha suggesting the same thing. Do projects. Yep. Christophe's think 63 percent is better than random guessing. Yep. And Christophe wants to know, what did you learn for the project? Sorry. What did you use for the project python? And so I could learn.

Paul: [00:32:36] Yes, I use Python and I just use like a UC Berkeley plugin to do all the data analysis

Harpreet: [00:32:43] And like table

Paul: [00:32:45] Manipulation. And then I kind of just built the K nearest neighbors from scratch, just like based off of a lab that I was doing in

Harpreet: [00:32:55] The Data science

Paul: [00:32:56] Class. So like we did a lab about K nearest neighbors, and it was about like predicting the genre of the movie. So then I just took everything in there and then just did it with these pickup lines instead. So, yeah.

Harpreet: [00:33:13] Ustin go for it.

Austin: [00:33:14] Yeah, I've have topologies, and I don't know technically how this would work exactly, but once if if you're sort of worried about industry standard and and using something like a like the sort of Berkely system for this, I think one thing you could do is if it's an interesting project to you that you want to pursue or push a little further, you could sort of reproduce it using

Harpreet: [00:33:34] Sort of more

Austin: [00:33:35] Tools like psychic learner or like that are out there that the industry is using. And so that sort of can like help you become more comfortable with those and actually show them in a very unique project through those tools. And you've already done a lot of the heavy lifting up front to like set up the project and, you know, come up with the approach and everything like that. And then the last thing I wanted to say, too, is like, you know, I think I a [00:34:00] comment we really need to surround around of hiring of like data scientists and machine learners. And the thing that was more important was like the communication around process and less around like, you know, given, you know, a sample data, how accurate were you able to train a model? It's more around like how do you communicate that? How do you communicate the limitations of your approach? So if you're on this first version, it's projects like what were the next steps would be we would have to experiment with like a do a run hyperparameters sweep or we would have to experiment with X, Y, Z, different algorithms to to, you know, improve on this baseline. And if if, you know, if hiring managers or whoever.

Harpreet: [00:34:36] See that

Austin: [00:34:37] They see that you have this sort of like regimented, thorough approach from going

Harpreet: [00:34:41] From a simple

Austin: [00:34:42] Test like baseline model. And you understand how to add complexity along the way. And not just like the first result was 99 percent accuracy. Like that's not the most important thing I got to believe for for for most folks. So just just a couple of thoughts there. I think reproducing that experiment, using more common tools and then sort of like really leaning into that communication and expressing the limitations and being OK with that, I think is is super important.

Harpreet: [00:35:07] Yeah. Excellent skills to have, like as a professional data

Harpreet: [00:35:12] Scientist as well. Might as well start

Harpreet: [00:35:13] Developing those now that you just are streets ahead of your competition.

Paul: [00:35:20] Yeah, I think I'm going to open this project up and try to try it again, because I'm taking a class where we're learning more about machine learning, but using like pandas and like numpad and all the just like industry standards. So I think I'll just try to, like you said, just like transferred onto those instead. Yeah. Thank you, Austin. Yeah.

Harpreet: [00:35:45] And there's a link right there in the chat forat for the article on baselines A.. A. Asain reproducibly a

Harpreet: [00:35:53] Project whether someone

Harpreet: [00:35:55] Else is translating from Python to

Harpreet: [00:35:56] R or your own, an

Harpreet: [00:35:58] Old project that you now [00:36:00] know how to do better have been the ones where I feel I've learned the most. 100 percent agree with that. Revisioning something and just try to make it better is definitely a good way to build your intuition and feel better about how far we've come since you first started it.

Harpreet: [00:36:17] Shout out to

Harpreet: [00:36:17] Everybody else that joined in the

Harpreet: [00:36:18] Room. A lot of new names. Well, there were

Harpreet: [00:36:21] Some

Harpreet: [00:36:21] Names, but they

Harpreet: [00:36:22] Dropped out. All right. How's it going? Natasha has to go in Bobba one. How's it going? Biology, Joshua?

Harpreet: [00:36:28] What's going on?

Harpreet: [00:36:29] But if you guys have questions, feel free to let us know. Looking into the chat here on LinkedIn. By the way, folks tuning on LinkedIn, there's like 20 of you watching. By all

Harpreet: [00:36:40] Means. Let me know if you guys

Harpreet: [00:36:42] Have questions or just click on that link to come and join us in the room. Mark is asking, what is your preferred platform for model deployment in both cloud and hybrid environment?

Harpreet: [00:36:53] It's going to be whatever the

Harpreet: [00:36:55] Company that I'm going to

Harpreet: [00:36:57] Work at

Harpreet: [00:36:57] Is using. Yes, that would be that'd be the way I'd answer that. Um.

Kawkab: [00:37:03] Yes.

Harpreet: [00:37:04] You want to come into

Harpreet: [00:37:05] That to chat, and that takes us little bit more in depth.

Harpreet: [00:37:08] Mark definitely loved it.

Harpreet: [00:37:12] But yeah, that's obsess over the company is already using. But if you do something just

Harpreet: [00:37:18] On your own. I hear I hear using A.W.

Harpreet: [00:37:21] As is fairly easy. So is GCP.

Harpreet: [00:37:25] Things that are a

Harpreet: [00:37:25] Little bit more easier to use

Harpreet: [00:37:27] Than Azure?

Harpreet: [00:37:28] In my experience, I've battled with both. I just felt like the documentation on AWB was much more neater and cleaner than the Azure Microsoft documentation was. Not that great. Very hard to do stuff with. So hopefully that answers your question mark, even though it was not really an answer. Let's keep a go and see if anybody else has questions.

Harpreet: [00:37:47] The best

Harpreet: [00:37:47] Questions. Let me know. Ok, anything coming in from YouTube or LinkedIn?

Harpreet: [00:37:51] Or is it the

Harpreet: [00:37:51] Building, too?

Harpreet: [00:37:52] I just thought. Oh, snap.

Harpreet: [00:37:53] Good to see you, man.

Harpreet: [00:37:55] Staff has Cristoff.

Harpreet: [00:37:58] You typed a question in somewhere. Let's see if [00:38:00] you are if your microphone is working or not. Let's give it the third time is a charm.

Krzysztof: [00:38:06] Can you hear me?

Harpreet: [00:38:07] Yes.

Krzysztof: [00:38:10] I changed my mind. My question was about the reach out. And you mentioned at the beginning, again, this topic of burnout.

Harpreet: [00:38:23] And I

Krzysztof: [00:38:23] Believe this is pretty important to have something outside of the work that

Harpreet: [00:38:29] Let

Krzysztof: [00:38:29] You clear the mind and everything. So I'm just interested to hear from you. How do you guys recharge to take the most of this time

Harpreet: [00:38:40] Between work

Krzysztof: [00:38:41] And work?

Harpreet: [00:38:43] Yeah, man, I wish I wish I had a good answer to that. I wish I had like actual hobbies that were not just learning stuff like be charting is just I suppose to my way of

Harpreet: [00:38:55] Recharging is typically man like I live

Harpreet: [00:38:56] A very regimented type of of life. Reik, I like it, you know. Forty eight a.m. do my thing, then go for a walk and then be extremely regimented.

Harpreet: [00:39:06] And my way

Harpreet: [00:39:07] Of recharging is just to completely destroy the routine and just

Harpreet: [00:39:10] Live how I want. Right. And I've been doing

Harpreet: [00:39:12] That for like the last last few weeks, really since I've been off work. And that's been helpful for me.

Harpreet: [00:39:20] Just just, you know,

Harpreet: [00:39:21] Removing all pressures from myself and just waking up when I want to

Harpreet: [00:39:25] Sleep like, you know, waking up and sleeping in.

Harpreet: [00:39:29] Now, I don't say what I want because it's very fluid on how I do things now. So I guess that's one way I recharge, is just completely destroy

Harpreet: [00:39:37] All of my routines

Harpreet: [00:39:38] And regimented lifestyle and build a back up. I find I do that

Harpreet: [00:39:43] Pretty often when I need to

Harpreet: [00:39:45] Recharge.

Harpreet: [00:39:46] It might just be like a week in,

Harpreet: [00:39:47] Week out, a month, or just could just stop being so regimented. And it's built back up. That's that's my way of recharging is just destroying and then coming back. I'd love to hear what other people do.

Harpreet: [00:39:59] Austin. Are you? [00:40:00]

Austin: [00:40:00] Yeah, that is a good question. I think I like Tour's answer. Just do something different, try new things. And I think to even expand on that is like talk to people who do different stuff and have a different set of concerns and a different set of things that are making them struggle. And I feel very fortunate that I'm in a wonderful relationship with my partner who is just ready to engage and all those things. And I find that sometimes like trying to help her kind of work through the things she's struggling with, gives me energy to like go back and sort of solve the own my own problems and I'm dealing with. And then I would say this, like anything you can do to kind of dissolve your like dissolve your ego a little bit like and I say that just like in the in the not in like your egotistical. I just mean like in the sense of like the

Harpreet: [00:40:49] Self thing that

Austin: [00:40:50] Just like creates all these like constraints that are sort of artificial on you. So anything you can do to kind of just like dissolve that, whether

Harpreet: [00:40:59] It's like

Austin: [00:41:00] And I mean, like reading like I think anticipating that reading fiction, just like anything that sort of takes you out of yourself, because I think like the feeling of being just so drained as you just I feel I feel that when I'm so internally focused and just like spinning over my problems that I can't solve over and over and over and over again. So whatever I can do to sort of just like dissolve the ego, some of the some, you know, some ways of doing that or I think are healthier than others, obviously. It's a subtle distinction from like escapism. I'm not suggesting escapism necessarily. It's more of just like figuring out ways to put yourself in the context of others, your surroundings, something outside of yourself is like super important for me to see that, like, oh, yeah, shit, this world is actually just bigger than like did my little fucking desk here that I sit at eight hours a day or ten hours a day or whatever it is, like whatever it is that that makes sense for you to do. That I think is is helpful for me at least.

Harpreet: [00:41:50] Absolutely love that.

Harpreet: [00:41:52] That's the kind of

Harpreet: [00:41:53] Very much in line with what I was saying when I'm at like kind of destroying the routine

Harpreet: [00:41:58] And like disintegrating all [00:42:00]

Harpreet: [00:42:00] That and just building it back up. I like. A lot. What about you, man? And then if everybody else would like to jump in here, let me know.

Harpreet: [00:42:08] Or go for it.

Harpreet: [00:42:09] Recharging, joining.

Harpreet: [00:42:10] This group has been

Harpreet: [00:42:12] Recharging for me. I mean, in the sense that it's like I said in my comment, it's just way out of my comfort zone. This is a complete different field that I and all of these things. But so it kind of gives me that other input. And sometimes I find it very relevant to what I do. And if it doesn't, then it's still relevant because I'm learning something. So that's one way. The other way I go and really recharge. Fully recharging is taking my old car and go for a long drive and just blasting music. And, you know, it's really to disconnect. I'm fortunate. I have a brain that works 24/7. So for me, music, watching TV, whenever it is just to stop the brain from

Harpreet: [00:42:59] Working

Harpreet: [00:42:59] On problems or anything like that, that's the key. And asked me after watching a TV show when I watched, I will not be able to give you an answer because it's just images passing by, keeping the brain, processing something that's not work related or projects or other things they've got going on at any given time.

Harpreet: [00:43:20] I love that, too. And that actually, I love driving. And so there's, you know, a couple of times where I'll just hop in the car and just drive around just to listen to music and just, you know,

Harpreet: [00:43:32] Go fast as well. Yeah, I like that. I like that a lot.

Harpreet: [00:43:36] And he likes to run do some cleaning, which is what he's doing right now. He says Ante's or rather, Austin says the word recharge kind of suggests plugging back in somewhere.

Harpreet: [00:43:47] But I can often

Harpreet: [00:43:47] Be breaking

Harpreet: [00:43:48] Those breaking

Harpreet: [00:43:49] Down those connections, allowing yourself the space to create new associations, etc. I like that a lot, too.

Harpreet: [00:43:55] There's a point

Harpreet: [00:43:57] About cleaning up, though, like Gordon [00:44:00] Peterson talks about this, like just clean up your room, just clean up your room. And, you know, it's like

Harpreet: [00:44:04] A part of the

Harpreet: [00:44:05] Cosmos that you have control over that you can come home to. They can come to grips with this like your space. You can make a beautiful you could do what you want with it.

Harpreet: [00:44:13] And it's quite helpful as well. I've actually done

Harpreet: [00:44:16] That, too, as

Harpreet: [00:44:17] Well as just started doing some

Harpreet: [00:44:19] Dishes.

Harpreet: [00:44:19] Right, plug in some

Harpreet: [00:44:21] Earphones and listen to something to do some dishes and

Harpreet: [00:44:24] And get right into it. Mark is

Harpreet: [00:44:26] Joining us at

Harpreet: [00:44:27] The right moment.

Harpreet: [00:44:28] Mark, what's going on? And Mark would love to hear your question. So your response to this question

Harpreet: [00:44:35] From Christophe Cristoff,

Harpreet: [00:44:36] Who is asking this, what do you do to recharge?

Harpreet: [00:44:41] My my favorite thing right now, my my job for like swag gave us a hammock like like a nylon hammock. And at first I was like, this is random, but I buy like like I use hammock or amicability for ham is like like 30 bucks. And like chilling on the hammock with like reading a book. The most relaxing thing ever. So like if the weather is nice, that's like what I'm doing at five o'clock. That's like the first

Harpreet: [00:45:08] Thing I'm doing. I like that

Harpreet: [00:45:11] Until we get some Covid Havocs, I'd would be that'd be awesome.

Harpreet: [00:45:15] I love that.

Austin: [00:45:16] We definitely look into it. I think that would be a very popular one, for sure.

Harpreet: [00:45:21] A.s. It's actually more important now that we're working from home, I feel. Yes, being able to recharge.

Harpreet: [00:45:26] Definitely. Man is weird, man. Like I was

Harpreet: [00:45:30] Just thinking about how

Harpreet: [00:45:32] Over the last year and a

Harpreet: [00:45:33] Half or so, on average, the furthest I've been away from my home, on average, it's been like two kilometers. Most crazy man that is like the same few kilometers

Harpreet: [00:45:45] Like it

Harpreet: [00:45:46] Gets Graney man. Like switching up the environment, trying to see something different is always fun. I think that's really helpful as well. I used to live by this

Harpreet: [00:45:52] Motto back in grad school,

Harpreet: [00:45:54] And it was that what new experience

Harpreet: [00:45:56] Can I have today?

Harpreet: [00:45:58] And I feel like I haven't lived up [00:46:00]

Harpreet: [00:46:00] To that in a in a long time. But I think I must

Harpreet: [00:46:02] Start reincorporating that like what? What can I see that's different? What's something new I can experience with? New song I can listen to was a new

Harpreet: [00:46:11] Youtube

Harpreet: [00:46:11] Video. I don't know whatever. Just something you experience up the new stuff. What about you, man? What do you do to recharge? And I want

Harpreet: [00:46:20] Sport when I have got time. I play I don't play

Krzysztof: [00:46:25] Anymore, actually, because of those restrictions. And because I wake up for my training Saturday night to 11 PM, so it's either training or sleep. So I try to sleep. And I love

Harpreet: [00:46:45] It's like really

Krzysztof: [00:46:46] I love spending time with my daughter because she's for us or for adults. It's amazing how quickly kids learn. And I just admire everything she does and. She's she wants to help with everything. By the way, she's 17 months, not right now. So she's got the management. But she's she loves everything we do. She she just wants to copy everything. That doesn't matter when we do. Any little thing that we hate because I mean, like cleaning or anything that we postpone.

Harpreet: [00:47:31] She she enjoys it.

Krzysztof: [00:47:32] And that's giving me like a new look to everything.

Harpreet: [00:47:36] And it helps

Krzysztof: [00:47:38] Me also to clear my mind. I mean, there's a time when I don't think about the things they're reading and stuff. And that's my thing

Harpreet: [00:47:48] Extensively about my son's like six months, just right around the same age as born.

Krzysztof: [00:47:53] He was born I think there was like seven weeks difference. I don't think I would think I'd buy Data the [00:48:00] twenty first marriage.

Harpreet: [00:48:02] Okay. Yeah. Yeah.

Harpreet: [00:48:03] Okay. So I was very close. My son was born May 8th. Did they did they learned super equipment is so fascinating and so that, you know, when I see him learn, like I can't help but think of like machine learning or artificial intelligence, just like how does he learn so quickly?

Harpreet: [00:48:17] Like, I could point to one thing. Right.

Harpreet: [00:48:20] I can point to one car. He knows what one car looks like. Now, all of a sudden, he knows that everything that looks like this is a car. It's interesting because he's made the connection that I should start calling objects. I should give them names that people know that I'm trying to communicate, that I want this objects that like some random things, like he calls those soother. He made it. He made up a word for Sudie that we've never heard. And he just referred to it by this this this thing that he calls you, because he goes through this a bit that you have no clue where you came up with that. He's got this he's got this picture frame

Harpreet: [00:48:55] In his bed in

Harpreet: [00:48:57] A bedroom, and he calls that picture frame Google. So how do you come up with these names and you just able to able to realize that I should name objects

Harpreet: [00:49:07] Or at least refer to

Harpreet: [00:49:08] Objects by me like just me? That's just mind boggling to me. I wonder if there's any brain scientist or neuroscientists that are listening to help help explain that to me, because I found that to be

Harpreet: [00:49:17] Extremely, extremely

Harpreet: [00:49:19] Fascinating.

Harpreet: [00:49:20] But then I also a

Harpreet: [00:49:22] I'll put on like some great courses and I'll make him sit down and

Harpreet: [00:49:25] Watch geometry

Harpreet: [00:49:26] With me. And he seems to really enjoy it. He'll just be sitting there chilling, like, well, just stuff like writing and writing some lectures on geometry and multivariate calculus and enjoy that sort of thing. He's not like my son. Let's keep going ask questions on anything whatsoever. Got a good audience here on LinkedIn. Had a had a bunch of people joining in. We'd love to have your questions. And also here in the chat, bunch of people that are here but haven't have participated

Harpreet: [00:49:54] Would love to have you guys up in anything whatsoever.

Harpreet: [00:49:57] Guys, I'm going to jump in today because I'm struggling [00:50:00] at the moment. I started playing around with CRYPTO'S. And, you know, I'm in a big money just a little bit here and there. And I've been doing it now for about a month and a half. And I cannot

Harpreet: [00:50:11] Find any trends or

Harpreet: [00:50:16] I can't understand what's going on. You know, with shares and companies, you can kind of have a feeling because it's so established. But for those, all I'm seeing is just ups and downs, ups and downs and one minute to just jump 30 percent. And I'm wondering, is there a way that you could use machine learning, for example, to kind of go back and look at the historical data to kind of give us some idea to explain what's going on? I mean, I have my own personal theories based on my past four weeks where I'm seeing that it seems to me that the sun days have a tendency of seeing quite large increases. And I'm feeling rather sitting at home there on the phone or laptop, and then they bark. Fridays have a tendency to see drops in the evening, especially it rises during the day, but it has a tendency to fall into the evening. Some figured probably somebody sitting at the bar selling off some candles so that they can pay for the beer. I mean, there's

Harpreet: [00:51:16] In a funny

Harpreet: [00:51:16] Way. So I'm just wondering if there's a way to take that historical data use machine learning to kind of see if there are a time relationship. Relation between time of day, weekdays, et cetera, versus the ups and downs of the CRYPTO'S versus the main ones, which is the Bitcoin and the after. I hope this topic can be relevant, but

Harpreet: [00:51:41] That's a hard question. I mean, question, if you knew the answer to men would be millionaires.

Harpreet: [00:51:47] Well, that's what I'm trying to figure out, you know.

Harpreet: [00:51:49] Yeah, I wish I wish Carlos

Harpreet: [00:51:51] Around my wife.

Harpreet: [00:51:52] Decoyed would jump in, jump into Cristoff here. But I mean, I'd say I don't like with [00:52:00] regular markets.

Harpreet: [00:52:01] Right. They have

Harpreet: [00:52:02] An opening bell and the

Harpreet: [00:52:03] Closing bell. Right.

Harpreet: [00:52:05] So it might make sense to to apply some type of time model to those prices. But cryptos

Harpreet: [00:52:12] Like

Harpreet: [00:52:12] 24/7. There's no market for it. Right. So you might need to bring in if you

Harpreet: [00:52:18] Wanted to understand

Harpreet: [00:52:19] How it's there and things are affecting that. Yeah, you probably would need to look at like

Harpreet: [00:52:24] Data, I'd say,

Harpreet: [00:52:25] Like, you know, scraping credit or something and getting some type of uh. And I just get

Harpreet: [00:52:30] That that

Harpreet: [00:52:32] I just get the feeling that this kind of like two things going on and the Clippers kind of fresh and new. So it's mostly emotions that are running it. That's like what people see on Facebook or read in the newspaper, which is kind of like the ups and downs, like suddenly the Bitcoin takes off and then everybody's buying bitcoins. And. But I think there are a few

Harpreet: [00:52:54] People that are

Harpreet: [00:52:55] Doing this more professionally in the sense that they have some kind of I'm thinking, you know, if there was a learning algorithm or some sort that you could say that if it increases 20 percent, you sell 200 percent of your clients. And then if you see a drop of 20 percent, you buy 40 percent. And to me, this kind of over time, I would like to test those things like those kind of what would what if. Ok, based on historical data, if I was putting in these

Harpreet: [00:53:27] Parameters and could I

Harpreet: [00:53:29] Build that into namel, then I could then actually

Harpreet: [00:53:32] Play with that

Harpreet: [00:53:33] To give me some. Ok, this will happen if I have done this over the past. But instead of me sitting and thinking up this, is there a malfunction? I could actually do that.

Harpreet: [00:53:42] I recommend reading the book. It's called Machine Learning for my Use a Machine Learning or Deep Learning for Algorithmic Trading by Stefan Jansen. I haven't gone through the book in its entirety, but I know it covers a lot of the

Harpreet: [00:53:55] Type of topics that it seems

Harpreet: [00:53:57] Like you might be interested in. Yeah, I mean, definitely, Kobad. [00:54:00] I see Kristof has his hands up. So if you have some insight on

Harpreet: [00:54:02] This, I'd love to hear it.

Harpreet: [00:54:04] I unfortunately do not. Cassandre from from LinkedIn here says

Harpreet: [00:54:09] It's all luck like like gambling that

Harpreet: [00:54:13] That's how I look at it now. It's it's gambling, you know, it's like I don't have. And that's why I'm just playing with little things right now. I'm just doing the little numbers to buy one or two coins because I want to try and get a feel for what's going on before you actually start thinking or to make it more of the potential. To me, it's like by a thousand different coins is just one of them takes off and gets into the Bitcoin range.

Harpreet: [00:54:43] You're OK even though you

Harpreet: [00:54:44] Lose them, all the others. But to do that, you need to kind of be the beginning. And now is the time in my money. And over time, it's going to be established like everything else. So it's just to kind of try and make it happen now.

Harpreet: [00:54:58] So some some I guess, some fun, some that fundamental, but some supplemental reading. Nassim Taleb is looking at SPOK

Harpreet: [00:55:08] Right now, fooled by

Harpreet: [00:55:09] Randomness, just to kind of paint the uh, to add some color commentary and to like, you know, how he thinks about trading and how he views. Random is a great book. And another one was a is a book by Ben Wall Mandelbrot. Where is it? The misbehavior of markets have been Wall Mandelbrot. Um, I haven't read into that one yet, but I will soon. I plan on going deep into algorithmic trading with deep learning in the very near future so you can keep an eye out for some content around that. But Christoph Ghafoor, I know you had your hand up.

Krzysztof: [00:55:47] I was just thinking how many people who are like to advanced like professors and people with multiple years of experience in A.I. and deep learning and stuff. [00:56:00] How many of them are buying based on their models? And I think there are

Harpreet: [00:56:05] Very few, if any.

Krzysztof: [00:56:07] Because this doesn't work like that. I mean, you may find some trends. There is, of course, something like a time serious analysis. And you you do it with like different kind of deep models, like recurrent neural networks. There are those memories, Terance, like A.S.A. and stuff. But you can't predict the future. And you'd have to predict what Elon Musk is going to tweet tomorrow. Ah. And I just don't know. I don't believe that any people who do have experience and they could do it because they already have enough

Harpreet: [00:56:50] Knowledge that they do

Krzysztof: [00:56:51] It because they know it's not doable. That's my opinion about that.

Harpreet: [00:56:58] Really get comments coming in from Rodney here on LinkedIn. Rodney saying that machine learning is Bhatta explanation's. That's why all the debate around that's why there's all this debate around explainable

Harpreet: [00:57:08] A.i. that

Harpreet: [00:57:09] Crypto is a time series problem. It is extremely hard to beat around Walke. I agree with you on that one as well. The crypto market is largely speculative. Almost everyone is trading on noise.

Harpreet: [00:57:21] And that's what I agree with that. Let me give

Harpreet: [00:57:25] You the reason

Harpreet: [00:57:26] Why I got into it. For example, I mean, with all these new online banks, et cetera, I signed up for an online banking and now they're offering to buy changed crypto. Paypal is now introducing it. So basically it's now coming out to the general public. And in my mind, the general public, including me, has actually no clue. But yet I started by, you know, a little bit here and there. This, to me means that the demand will increase and it's going to increase rapidly in the future as these online banking services sector is going to start to offer [00:58:00] this service. And I mean, all the warnings that they don't invest, if you don't know, and blah, blah, blah, and that's fine. But, you know, we know how people are. It's like you said, the Twitters and Elon Musk and the latest news that's going to trigger. But the impact of those triggers are just going to be so great because now more people have access. I believe that Bitcoin previously it's more difficult because you have to have an exchange platform, an account in the beginning, and there was more with all of these.

Harpreet: [00:58:31] It would be good to get into.

Harpreet: [00:58:33] It was not easy. Now is becoming so easy that the volume is just going to be the driving force is going to be a driving force. That's what I believe. And that's kind of why I'm now kind of thinking that maybe it's

Harpreet: [00:58:47] Not because I

Harpreet: [00:58:48] Believe that Bitcoin makes sense or any of these coins actually has some. It's not like a company. We have a shared value that the coins are more feliks like the guidelines. So this is why I wouldn't mind to see if I could find a project and find something that can make this program. And I could do the analysis, too.

Harpreet: [00:59:07] Yeah, I think the more interesting thing is actually just the block chain technology itself.

Harpreet: [00:59:11] Right. Like, you know, Bitcoin is just

Harpreet: [00:59:13] One implementation of blog chambre, just the block chain technology itself, like the applications that are far ranging, that it can have effects on the feature that we can't even imagine. Right. Like Bitcoin can just be like, I'm dating myself here, but Bitcoin can just be the

Harpreet: [00:59:27] Friendster right now,

Harpreet: [00:59:29] Right off social media. Right. It's before there was Facebook and all that stuff. There's like Friendster, MySpace, whatever. Right.

Harpreet: [00:59:35] Um, so

Harpreet: [00:59:37] But social

Harpreet: [00:59:37] Media as a as

Harpreet: [00:59:39] A thing is the the larger trend that, you know, be paying attention to. Kind of the same analogy I want to

Harpreet: [00:59:45] Make for blogs. Right. Like Bitcoin is just

Harpreet: [00:59:47] One implementation of it. But the technology itself is what we should be paying attention to. Mark. Go for it. And then after Mark A.F. fear and muted as

Harpreet: [00:59:56] Well before we jump

Harpreet: [00:59:57] In as well.

Harpreet: [00:59:58] After work. Go for it. Yeah.

Harpreet: [00:59:59] And that was [01:00:00] a that was a perfect Segway to the that Mapoon link to Carlos's book or decentralized finance and blog Chambray. Is it like a pretty good intro into understanding the space donation based? I believe so. It's it's a great book and fun to read, but essentially similar to what I was saying is like I think one of the main things is that you need to look at the underlying technology for these new coins. So it's it's like a lot less random in that sense where you're actually more so like approaching it like like an investor in a

Harpreet: [01:00:32] Way that an investor

Harpreet: [01:00:34] Has a consumer like investor in a business where you're seeing

Harpreet: [01:00:36] Like where the type

Harpreet: [01:00:37] Of contracts, how is this coin? Devi's like, what's this business utility used for for that so.

Harpreet: [01:00:46] For example,

Harpreet: [01:00:46] For like I'm not a block chain crypto experts and things of brand salt, but like Bitcoin, that was like one of the first market to show proof of concept like block chain work in a financial institution or a financial

Harpreet: [01:00:59] Tool. But then you have

Harpreet: [01:01:00] A theory and similar thing is like all around smart contracts and having a method in which you use crypto to set forth like contracts and actions execute automatically. And so like there's different types of like underlying technology, like applications and uses. They're like stable coins that match the value of the dollar, which has a utilization. And so I think it's like less about like what's going to be the next hot thing in the sense of like what's going to pop off. There's no way to really understand understand that this is like completely like noisey,

Harpreet: [01:01:34] But rather like what's the

Harpreet: [01:01:35] Underlying technology of why is this coin being built? What's its purpose?

Harpreet: [01:01:40] What is unique? What makes it unique

Harpreet: [01:01:42] For further adoption, among others saying the kind of talk about the random is like Dogecoin Doji going where HouseCalls. That's crazy Populaire. And it was created as a joke. Why even try? It's got Mehmed into existence, right. So there's no way to predict [01:02:00] that. And that has no utility really designed within it as compared like other coins.

Harpreet: [01:02:07] I think another one, and I think

Harpreet: [01:02:09] Carlos mentioned is that coin basic attention, some of the long lines where where essentially the coin is designed to pay creators say like, hey, I'm going to support you and give you this coin because I watch an ad in exchange.

Harpreet: [01:02:24] So there's like different technologies.

Harpreet: [01:02:26] And I think that's like a key thing. Like I wish a lot more people looked into for for crypto is like the technology applications of it beyond just being speculative. But the thing you know, it's like I'm lazy. I'm not a big book reader. I have to admit. So I've kind of, you know, try and experience abroad first. But I but I see your point. I agree with you. I've done some white paper reading just to get a little bit more behind. I got a few dodge coins. You never know. You know, it takes salt,

Harpreet: [01:02:57] But it doesn't have

Harpreet: [01:02:58] A proper value. When you read the articles from people that quote unquote understand the industry, they're all saying the Dogecoin. It's I think it's just emotions because it doesn't have there's no purpose. It doesn't have to you can't use it for anything at this moment. But then again, one day is maybe somebody figure a way

Harpreet: [01:03:19] To use it

Harpreet: [01:03:20] For something. Who knows?

Harpreet: [01:03:22] Yeah. Man, just excited to see where where these trends take off. And I think spending some time, a couple of hours, you know, a few hours that we use just to understand the bloc's new technology and to see different applications for it. And then maybe think about how those applications

Harpreet: [01:03:41] Can be

Harpreet: [01:03:42] Melded together with machine learning, I think would be very fruitful anyway. It's a good exercise to undertake to find those intersections, get in on them early, like Carlos is doing that right now. Right. He's on the you know, he's he's interested in it. And because he's interested, he's pushing so much energy and effort into it, he's understanding something. [01:04:00] And this is going to pay dividends to him. He's he's exploring something new and novel. He's on the frontier of it.

Harpreet: [01:04:07] You know, you got

Harpreet: [01:04:07] To be a little bit essentially to be around the frontier.

Harpreet: [01:04:10] But he's just going deep into it just because he's

Harpreet: [01:04:13] Interested and you got to do the same.

Harpreet: [01:04:16] I still think as well that Carlos, basically, if the

Harpreet: [01:04:19] Block chain, like for

Harpreet: [01:04:20] The bootcamp, he's doing me a post about how essentially create his own loan

Harpreet: [01:04:24] To pay for it based off

Harpreet: [01:04:25] His crypto assets, which is wild, which is like shore legalization of like crypto and block chain technology outside of like the financial aspect of itself, but like creating tools to mitigate the need for for like bots calling a trusted third

Harpreet: [01:04:40] Party. Yeah.

Harpreet: [01:04:41] Yeah. So great topic toward things for open up that discussion. See some comments coming in from from LinkedIn.

Harpreet: [01:04:48] Thank you,

Harpreet: [01:04:48] Cassandra, for your comments. Just saying that,

Harpreet: [01:04:50] You know, we can't

Harpreet: [01:04:51] Predict the future of

Harpreet: [01:04:52] Then we would be

Harpreet: [01:04:53] Able to stop natural disasters. Yes, that is true. Any other questions or comments? Please let me know. Shout out everybody in the room that we have not heard from. That's Natasha Bobba blog. She A..

Harpreet: [01:05:03] Joshua Pirat. If you guys

Harpreet: [01:05:05] Got questions,

Harpreet: [01:05:05] Now's the time. We are going to

Harpreet: [01:05:06] Start wrapping it up, OK? Mark has a question. Go for it.

Harpreet: [01:05:10] So my my good friends learning software engineering right now is in deep in it. And he asked me a question. He's like, what's the difference between writing code and writing production level code? And I kind of struggle to give an answer because I've kind of done my job and that kind of have done my job. But I'm curious what other people think.

Harpreet: [01:05:28] Like what? But what separates

Harpreet: [01:05:31] Production level code from others

Harpreet: [01:05:32] Think production level could

Harpreet: [01:05:34] Probably a lot more

Harpreet: [01:05:35] Thoroughly and rigorously

Harpreet: [01:05:37] Tested,

Harpreet: [01:05:38] Probably more of version control, because you've got multiple people working on

Harpreet: [01:05:42] The same chunk of

Harpreet: [01:05:43] Code, essentially the

Harpreet: [01:05:44] Same body of.

Harpreet: [01:05:46] Um, yeah, I, I probably the two biggest things, I guess more thoroughly tested, it's more Russian

Harpreet: [01:05:53] Controlled and

Harpreet: [01:05:56] More well documented for sure. You have to really outline [01:06:00] your thinking process. And I copious documentation.

Harpreet: [01:06:03] Um, but that I flip this one

Harpreet: [01:06:05] Over to Cristoff

Harpreet: [01:06:05] Since he's, you know, actual

Harpreet: [01:06:07] Software engineer. Uh, what

Krzysztof: [01:06:08] Do you think? I was afraid to hear that. I think it's just better. The design is better. I mean, there are

Harpreet: [01:06:21] I mean,

Krzysztof: [01:06:22] Everything what Harpreet

Harpreet: [01:06:23] Said, and it

Krzysztof: [01:06:25] Needs to meet some standards like good design. I don't have really so much experience of the done production for JavaScript, like for Fronton two years ago. But it was like with this react thing, which is I mean, when you called we've riadh, there are some standards you just need to follow to not to get lost in the in the project, insert itself in the code. And I've donoso Java.

Harpreet: [01:07:00] And it was

Krzysztof: [01:07:01] Like pretty much the same. You choose some tool like libraries and they just guide you and you make sure that everybody's paying attention to a clean code in this case. But I mean, if you just call it center,

Harpreet: [01:07:20] I like the point about the the same pattern. Like you're like, for example, if you're writing in Python, then make sure you're strictly adhering to like Pepé standards. But sorry. So go for it.

Krzysztof: [01:07:30] Um, I was actually at time, I mean, like when you code,

Harpreet: [01:07:34] You can you can write really

Krzysztof: [01:07:36] Dirty code that does what you need and nobody looks

Harpreet: [01:07:40] At it, but it

Krzysztof: [01:07:43] Makes what you expect.

Harpreet: [01:07:45] But it's mostly

Krzysztof: [01:07:46] Like on the part of of the code, if you're working on like a bigger project that has multiple,

Harpreet: [01:07:55] Multiple,

Krzysztof: [01:07:57] Many parts, you can do code dirty [01:08:00] Kuhr dirty. They're dirty there because you're not be able to connect the parts. But it's also designed. Yeah.

Harpreet: [01:08:11] I mean, just went to the pragmatic programmer, be like the number of book I'd point him to like whatever these guys say. Yeah, but right there, that's like the

Harpreet: [01:08:21] The the Bible

Harpreet: [01:08:22] For a production level code.

Harpreet: [01:08:24] It's the first book I sent him. Yeah. But it was interesting when you asked me like I know how to write it, but like explaining it to another person. I mean, trouble coming up with words. Yeah.

Harpreet: [01:08:36] Because I mean, when you're doing code for your own project, it's like, OK, well, you could you can make all the bad mistakes that you

Harpreet: [01:08:41] Want to make because you're

Harpreet: [01:08:42] The only one interacting with it.

Harpreet: [01:08:44] Right.

Harpreet: [01:08:44] But when you're working on a team, like it's important to follow those design patterns. Right. Because they're there for a reason. Just makes it easier to review. Right. It makes it'll make it easier for things to stick out for people as reading, you know, another book by Andy Hunt, which was pragmatic thinking and learning. And he's talking about the importance of how following good design patterns and following the design conventions, for

Harpreet: [01:09:08] Example, happy just

Harpreet: [01:09:10] Reduces the cognitive load for somebody who is reviewing your code. See, I could pull that up, but

Harpreet: [01:09:16] I'll be

Harpreet: [01:09:18] Hard right now. But I guess that's kind of the the main differences. Production level code is code that is not just for you, that you're just going to be part of a larger body of work with many other individuals. So it needs to be more clearly

Harpreet: [01:09:30] Written, well documented,

Harpreet: [01:09:33] And make sure you're adhering to whatever coding style is applicable for either your team or that language itself. Hopefully that was satisfactory. Shout out to all the software engineers who are just roasting you right now on LinkedIn and is asking what's the relationship between deep learning and Bitcoin technology?

Harpreet: [01:09:51] I don't know yet.

Harpreet: [01:09:53] You'd have to explore that. I'm sure there could be some connections, could be some applications. So definitely worth exploring questions, [01:10:00] last minute questions for anyone else.

Harpreet: [01:10:02] Please let me know either.

Harpreet: [01:10:04] Baloji, Pirate Natasha, Joshua A. The 20 or so people watching on

Harpreet: [01:10:08] Linkedin, let me know,

Harpreet: [01:10:10] Does not look like there are any other questions. Go ahead and call it a day then. Guys, remember, next Sunday will be. We're actually going to be

Harpreet: [01:10:19] Off next

Harpreet: [01:10:20] Sunday. Wife and

Harpreet: [01:10:21] I will be in Vancouver, B.C.

Harpreet: [01:10:24] Just chilling out a little bit, leaving the baby with

Harpreet: [01:10:27] Deer, with four

Harpreet: [01:10:28] Grandparents here. I think he'll be well taken care of. So we won't be we will not be here next Sunday, but possibly the Sunday after that. I think we have to get that schedule set up. Don't forget to join in on the happy hours on Friday, this Friday, I've got our community member Antonio.

Harpreet: [01:10:47] Taking over the role of host for Friday's happy hour session. So that should be fun. Definitely. Make sure you guys show up, support him. Come on. Some good questions. Keep the conversation lively and fun. Guests tune into the episode, I believe, with Jeff Li earlier this week. Check out the interview with Ken Jee as well on his podcast and let me know what you guys think. Thanks for joining us, guys. As usual, remember, you've got one life on the planet. Why not try to do some big? Cheers everyone!