HH80-06-05-22-02.mp3

Harpreet: [00:00:13] Let's go. What's up, everybody? Welcome. Welcome to the Plaza de la Science. Happy hour. It is Friday, May 6th, the first Friday in May. Can't believe this may already be flying by so quick at this point, but I'm.

Speaker2: [00:00:29] Happy to.

Harpreet: [00:00:30] Be here. Hopefully you guys.

Speaker2: [00:00:31] Got a chance to tune in to the podcast episode that I released today. Very, very special podcast episode. I interviewed Marcus Du, so toI, who is a professor at the University of Oxford. He's well known on the BBC.

Harpreet: [00:00:47] Written a couple of really, really.

Speaker2: [00:00:48] Great books. One of my favorite books, which is the Creativity Code, which is all about how deep learning is able to augment humans creativity and and.

Harpreet: [00:01:00] Human just innovation.

Speaker2: [00:01:02] Arrow For whatever reason, deep.

Speaker3: [00:01:04] Learning.

Harpreet: [00:01:04] Things get a lot of hate on LinkedIn specifically, and I don't understand why that is.

Speaker2: [00:01:09] Deep learning is awesome. You should learn it.

Harpreet: [00:01:13] Like people that are just hating on deep learning for no reason I think are just scared of things that they don't understand. People use don't learn it. It's amazing. Do you have to learn it? No. But if you're interested in it, please learn it. Deep learning the soap. Greg Colicchio is taking over as host today.

Speaker2: [00:01:30] Greg, thank you so much for doing this. I'm about to announce in just a couple of minutes I'm.

Harpreet: [00:01:34] Trying out here in office, we've got rented like this giant cottage for my son's second birthday. So I've got my parents, my wife's parents, my grandparents, all of us in one house. Luckily, it is a massive, massive cottage.

Speaker2: [00:01:49] So that everyone can have.

Harpreet: [00:01:50] Their own space, which is much needed when.

Speaker2: [00:01:54] You get as irritable as I do around parental figures. But it's [00:02:00] good to be here for for just a little bit. Greg, I'd like to turn the reins over to you, my friend. Thank you so much for taking over.

Harpreet: [00:02:07] I really appreciate it. You guys take care. Have a good rest of the afternoon, Greg. This is all yours, man. Cheers.

Speaker3: [00:02:14] You got it, man. Enjoy time with family. So, again, thank you for letting me host this week's Happy Hour. So I'll try my best to live up to your specialty. Right. So enjoy. How's it going, guys? How's it going? It is super great to be with you all. And I'm already enjoying the presence of the familiar faces. We got Auntie, Surg, I'll be shek and. Great to have you all here. I'm trying to pull up the screen for the live section. If anybody is participating, I'll be looking at that to filter out some questions and bring it to the team here. So hopefully we'll have more participation. But for now, a quick thing for you guys. How was your week? Did you work on something that was weird or did you work on everything that was expected? You know what? What can you tell me about the week so far and what you're expecting to do for the weekend to kind of either forget or reminisce what just happened to your life?

Speaker4: [00:03:25] I'll break. I'll break the ice. Yes. So this I'm a one hour midget. Know in the military, if you have less than 100 days left, they call you a two digit midget. And so I'm not only less than a one digit midget, I'm a one hour midget. One hour left in my current role. I start a new role next week with the super cool company. Really looking forward to it. I already got my new toys, but this week I was trying to solve some [00:04:00] problems with the last company, which I'm still on good terms with, good friends with, and it forced me to dove into Docker. I really hadn't learned it to the depth that I wanted to, and I found a great, inexpensive course for it, enjoying understanding it better. And then this weekend and my first Monday on the job, I'll be out of town, which they were very cool about. But my baby biological daughter is graduating from college, so we're flying all the way to the other side of our country to be with her for that. And that's me in a nutshell. But Docker is actually a lot easier than I thought it was. And Node is every bit as hard as I ever thought it was just getting installed, right and everything. It can be such a bear sometimes, but I was playing with Node inside Docker and wish the examples were different. So I'm eager to hear from someone else though.

Speaker3: [00:05:02] Yeah, that's interesting. And by the way, you know, congrats on your for your daughter. It's it must be super exciting and you're probably reminiscing a lot of the things that you've done when you were in college and kind of like, you know, I'm pretty sure you were there to to accompany her throughout this journey and you're probably telling her to the journey is just getting started, right? She has such a long life ahead of her and full of adventure and things like that. So that's true.

Speaker4: [00:05:30] I have young people all over the world that hang on my words, but not my legal kids. No, they're through listening to Dad. They'll listen to me when they get a little older.

Speaker2: [00:05:42] I'm sure.

Speaker3: [00:05:44] That's funny. Oh, cool. And you said you just started with Docker. So what makes it what makes it what did you think? It was a little bit hard for you.

Speaker4: [00:05:54] So I started learning it about five or six years ago, used it a while then [00:06:00] didn't need to use it. But the way I first learned it was like, throw, throw you in the deep end. Don't learn the simple steps. And now I just decided, No, I really need to know it better. There's some reasons I need to start using it and I am thrilled. It's it's a great concept. It's a great tool. It's it's like having the separation level of a virtual machine without needing all of the virtual machine. It gets its own file system, that container does. And you can have everything isolated from the rest of the file system on your computer that you you're running an image on inside a container. But that way I can build something for you, Greg, and you can install Docker and you can run the Docker image in a container and it'll work on your machine just like it did on mine. No problems. You won't have to worry about installing dependencies because they'll all be installed in the image, all almost as well as strong as me sharing a virtual machine image with you, but not near as big.

Speaker3: [00:07:13] So if I understand it well, like if you have like multiple containers on your computer, on your laptop, do they all share the same like ram for your from your laptop, the same memory of your laptop? You're all using the same resources of your laptop and they're limited to that capacity.

Speaker4: [00:07:31] Yeah. Cpus, GPUs, ram. Etc.. Kernel of the operating system. But where it starts to separate is that the actual file system, the containers get their own file system. That's the that's the first thing to really make sure it's isolated from the rest of your system. And it's a true shipping container. That's why their logo is a picture [00:08:00] of a whale with cargo containers on. It's perfect image, but it could be a actually a giant ship that carries those cargo containers instead. In fact, the course I'm taking, that's the image he uses. That's very apropos because it just gives you such good isolation from the rest of your system. And it's truly an isolated container that that you can ship applications in and run them without issues. That's why it's so popular.

Speaker3: [00:08:32] And I'm hearing you said also GPU. So it has the capability of going to online service providers to, well, get computing power outside of just your laptop, right?

Speaker4: [00:08:45] I don't know about that yet. I'm just saying the system that it runs on, it will use its resources. It just it stops at the file system level. It's got its own file system.

Speaker3: [00:08:58] After. Good, good. I can see it where it allows better, I guess. Communication or collaboration with fellow data scientists. You're sending them something on their laptop, they can work with it. Same environment and things like that.

Speaker2: [00:09:13] Anybody else?

Speaker5: [00:09:15] Yeah, I actually it's my first time here, by the way. I've been I saw I can't remember the gentleman's name that was here before, but I saw his YouTube conversation with Daniel two, and he had mentioned something about you guys had this happy hour, and maybe you'd be open for mentorship, you know? So that's why I'm here. I am, actually, I don't have a data science background. I'm actually a chemical engineer by background, but I am [00:10:00] trying to transition into data science. So I've been spending quite a bit of time just learning data science through different websites. One of them is Data Camp, and I just finished about the whole module, which was about 77 hours or so.

Speaker4: [00:10:22] Happy shit. Can I can I ask you a question? Sure. Did you ever do any empirical modeling?

Speaker5: [00:10:31] What is that?

Speaker4: [00:10:33] Okay, so where you're just fitting some math. So you have some measured output and you're just doing a fit, a curve fit for that to create a model.

Speaker5: [00:10:48] Yeah. I mean, I have done that in my past job as a as a chemical engineer. We deal with lots of plant data, like different data that is originating from well.

Speaker4: [00:11:00] Now. Hang on. So you've done some predictive modeling. Yep. And did you ever do a ANOVA? Analysis of.

Speaker5: [00:11:07] Variance? Analysis of variance, all sort of.

Speaker4: [00:11:09] You ever do design of experiments?

Speaker5: [00:11:11] Yep, all the time.

Speaker4: [00:11:13] Abby Scheck. I christen the a data scientist who might need to learn some new language terminology. That is.

Speaker2: [00:11:21] Okay.

Speaker5: [00:11:23] Yeah, I, you know, I threw out my chemical engineering career. I had to look at a lot of plant data, and we used to use this different software. It's called PY Process Book and we used to build dashboards on which equipment is going to go down, which is going to how it's operating. But you know, done predictive data modeling for past seven years in my career. But you know and now all all we are programing and I'm looking at all these, you know, all these new stuff and it's linear regression and, [00:12:00] you know, all these different things and it feels familiar. But, you know, coming in from a different background, it feels a little bit intimidating to to say that, oh, yeah, you know, because programing is not my forte. But I have worked closely. Like I said, I've been working, working on programing since past one year, not there yet, not an expert by any means, but would love to get some insight over what should I be doing, how do I break into this field, and what are the pitfalls? And, you know, how do you initiate the conversation within within my organization and I work for a bigger organization, big, big semiconductor company. And, you know, it's it's yeah. So that's where I am.

Speaker3: [00:12:54] Well, Abhishek, you and I have similar background. I'm an industrial engineer, but seems like based on time and a lot of here here in the comments here and the comment that is you've been you're a de facto scientist. So I think all of us here in this channel has had some sort of adventure with data where we've tried to model certain, I guess, pattern that we suspect we see in the data in in this modeling helps us either, you know, predict where what the final output would be. So I used to work for chemical company and one of my process engineer's colleagues he's he was it was this chemical company that was like, let's say grinding copper, right? You're putting it in a big tank and then you're grinding it to find powder in the liquid. And it turns out that copper is actually very potent to like bugs like the wood. So they treat wood with this concoction. [00:14:00] Right. And he would model how fast, how much time you would take to kind of pulverize these, you know, size these grains of copper to a certain size that is acceptable or small enough that can penetrate the wood material. So he's modeling. You call himself a data scientist, but the way he was modeling those things, I mean, I'm like, man, you're a data scientist. There's nothing else like you can tell me how long you will take to grind this thing and what is the final size of these samples and things like that. That's that's the work of a data scientist. And you can tell me what kind of vibration the machines have and that will predict a final size after X hours of work, etc., etc.. So, you know, do, do see an opportunity to bring your, I guess, your chemistry knowhow to this, to this realm of data scientists, and you'd be a much more powerful colleague for your friends. So I know you have something to add.

Harpreet: [00:15:09] You're muted. I'm muted. Yeah. I just wanted to say welcome to the field. I definitely agree that you're already a de facto data scientist. I mean, there's so many people that perhaps came from come from a CC background, maybe, you know, worked as a software engineer, never dealt with data. They take one Python course and they think they're already a data scientist because they can do models that fit. And that's definitely not it. You're more of a data scientist than they are or ever will be. I mean, you just if you learn Python or R, I mean, that's it. You have all the tools you need.

Speaker5: [00:15:53] Thank you for saying that. Makes me very gave me that energy that I needed, I guess. [00:16:00] You know, I have a couple of different mentors at my company, too, and they all say, like, you know, there is an imposter syndrome that that I tend to suffer from. And, you know, I've been trying very hard, like past month or so. I've been giving close to 5 to 6 hours every day, just grinding through going through that data camp, you know, learning bash, op MongoDB and all all sort of stuff that they had in the data camp. And, you know, I still feel like I don't I don't know what else do I need to do. So I even got my I'm actually going to starting my master's here in artificial intelligence coming fall, you know, but I've talked to a lot of different people. I even doing a short of an internship within my company as a data engineer. But how do you I guess my question is, you know, I saw I see one of the questions. What are you wanting to do that you don't already? I do want to get into like a pure data scientist role. And I was just thinking about like, what else should I be doing? How do you get good at it? Because I one of the problems that I see that I don't remember all the codes. Right. I write it today and tomorrow. I don't remember anything.

Harpreet: [00:17:37] It's crazy. That's what StackOverflow is for.

Speaker5: [00:17:40] Yeah, I don't know. It's a new programmer problem is.

Harpreet: [00:17:47] Every programmer problem.

Speaker2: [00:17:48] Okay.

Speaker3: [00:17:50] Maybe. Maybe he's the approach that matters the most, right? Eric, I know you probably have some good insights on that, so I'd like to hear from you. [00:18:00]

Speaker2: [00:18:00] Instead of a follow up question for you. So what do you want to do that you're not doing? Already said you want to be in a pure data scientist role. What does a pure data scientist role look like to you?

Speaker5: [00:18:15] A pure data scientist role would look like someone who is delving dealing with data on a daily basis, someone who could model, build dashboards, you know, wrangling data. So right now, I'm working as a project manager, project engineer, as a manufacturing project engineer. So I'm mostly dealing with like building, you know, building physical plants because that's my daily job. And, you know, the reason one of the reasons why I wanted to get into this field is because I like the remote aspect of it. And, you know, with the COVID, everything has changed. But they have now started calling us back, you know, just because I have to deal with like physical stuff, you know. So yeah. So just if I could, if I could do a little bit more with data, I think I would be. I think that's where I want to be at.

Speaker2: [00:19:18] At it. So just just to like I can speak from where I work, so where I work. People who build dashboards, for example, are not the same. People who build ETL pipelines, who are not the same, people who build models. And so yeah. And so if you are looking for a data engineer role or a data scientist role or a data analyst role or reporting analyst role or, you know, and so that's I was just as you kind of as you kind of said that like in that case, pure data science just means based on what you said and what I kind of process was like, it's like kind of like having data in the title. But then from there, there's a whole world of like different, of [00:20:00] different roles. Is that kind of what you're my understanding correctly?

Speaker5: [00:20:03] Yeah. No, I think, you know. Sorry, I misunderstood your question. The end goal is to become a machine learning engineer. That's. That's where I want to go. Yep. So. But, you know, when I was doing my gig at this organization or sort of an internship, they had mentioned that, you know, most data scientists, the quality of the data that we get is not good. So you have to start with you first need to understand where the data is coming from and if it's clean and if if it's clean, then you can go do other things. So the hard part about data science or any machine learning is having good data. And I wanted to start at the very beginning in the ingestion phase and and transformation doing ETL. And then once I get good at it, maybe I could transition into a little bit more data science and then machine learning and all that because I figured, you know, I had the math and it's that background. I understand the concepts that the general machine learning talks about. I just don't know the models yet, but I have a book here in front of me. You know, it doesn't show up, but it's the it's the hands on machine learning with this guy could learn. And TensorFlow, it's the Bible run. I think everybody has this book. So I've been going through that, reading through that and going through YouTube. And that's that's how I've been learning. So.

Speaker3: [00:21:47] Tom, I know you have an ambition, but I want to acknowledge people who may be watching. Following directly through LinkedIn. I know Cara Wicks may have mispronounced [00:22:00] the last name here said to a comment that Tom said previously about Abhishek being a data scientist without him knowing he's a data scientist. Is that whatever Tom said, well stated language can be so funny. Different words mean essentially the same thing a lot of times. Definitely learning is coming from a background in experimental psychology. With that, I have a question later on about psychology and the role we may play in this big field of data scientists. And what do you think this is going next in the next ten years? But, Tom, I'd like to hear from you.

Speaker4: [00:22:35] Just very quickly. Abhishek, I don't know if you've opened the chat yet. We were encouraging you in there too, but I put my LinkedIn profile in there. I'd love for you to connect with me. I tried to find you, but wasn't sure what Abhishek Shah you were. That's a more common name than you know. So please connect with me and then I'll make sure you get connected with most of these people too. And we have a we have a chat group that will make you part of so we can help you any time you need. But you're I assure you, it's not going to be hard. It's just more language and new techniques and relating it to what you already know. You'll you're going to love it.

Speaker5: [00:23:19] Just I actually already follow you on on LinkedIn. I didn't, you know, following you for a very long time, actually.

Speaker4: [00:23:26] Oh, good. I'm talking about being connected, though, and then we'll add you to that check or most of us here are in it. And if someone wants to be in it that's here today, we're not trying to keep you out of it. Just let us know. It's a LinkedIn chat group.

Speaker3: [00:23:42] But also want to acknowledge something that Russell said in the chat. And under Russell, if you can go, if your audio works now, if you want to read it or restate it and you want to read.

Speaker2: [00:23:57] It for me, I think. Can you hear me?

Speaker3: [00:24:00] Yeah, [00:24:00] I can hear you. Yeah.

Speaker2: [00:24:01] Apologies, everyone. My laptop did a big windows update earlier in the week and it just killed everything.

Speaker3: [00:24:07] So my audio's just about.

Speaker2: [00:24:08] Working, but I don't think I'm going to.

Speaker3: [00:24:10] Get a video.

Speaker2: [00:24:11] Today. So just to reiterate what I what I was saying in response to.

Speaker3: [00:24:15] Your description.

Speaker2: [00:24:17] To Eric's question, Abhishek was I heard you say that you were keen to make dashboards and reports. So my initial response to that was if you're if you're keen to build those types of things and perhaps a data engineer.

Speaker3: [00:24:31] Direction would be better than.

Speaker2: [00:24:32] Pure data science. For me, data.

Speaker3: [00:24:34] Science is kind of really getting into.

Speaker2: [00:24:36] The weeds, acquiring the data, cleaning the data and transforming the data.

Speaker3: [00:24:41] The data engineer does some of that, but then transitions that.

Speaker2: [00:24:44] Through to the to the end result in a report and then transitions that into.

Speaker3: [00:24:51] A data analyst. So for me, the data engineer is kind of a good halfway house.

Speaker2: [00:24:55] Between the data science.

Speaker3: [00:24:57] And the actual report production.

Speaker2: [00:25:00] So that might be a good alternative route for you to consider. Although you then subsequently went on to say that ML engineering.

Speaker3: [00:25:07] Is probably your ultimate goal.

Speaker5: [00:25:09] Yeah, that's where I want to be. But you know, I have read, you know, again, all of my knowledge is sort of like getting it from reading LinkedIn profile to getting Internet. And a lot of people will say like, okay, you should not transition into an ML engineer right away. You should try to get into data science first and see how you feel about it and then you can become an ML engineer. I don't know how much of that is true, but you know, ultimately I guess I'm really interested to become a machine learning engineer.

Speaker2: [00:25:47] Sure. Sure. I know that sounds like a great goal to have, and I would support your slight skepticism about what you're seeing on social media. You know, take take all of that with a pinch [00:26:00] of salt. Don't let it dissuade you from something that you have a passion for. That being said, I think perhaps there is some truth in the you know, don't jump into mal engineering, you know, two footed with a blindfold on.

Speaker3: [00:26:12] Get some knowledge in other fields first. But, you know, maybe don't spend ten years doing data science before you go to engineering.

Speaker2: [00:26:19] But have a have a good appreciation of a lot of the other fields that relate to that. But yeah, engineering is.

Speaker3: [00:26:27] Well, ML engineering and ML ops are two.

Harpreet: [00:26:29] Really.

Speaker2: [00:26:32] Zeitgeist kind of.

Speaker3: [00:26:33] Roles at the moment. So you will almost certainly see a lot of that on.

Speaker2: [00:26:37] Linkedin.

Speaker3: [00:26:38] Especially and perhaps other social.

Speaker4: [00:26:39] Media as well.

Speaker3: [00:26:40] And if you're if you're confident that you know what that.

Speaker2: [00:26:44] Is and you want to go for it, then yeah, go for it. But as I say, get a taste in.

Speaker3: [00:26:49] Appreciation of other roles that work with it as well, so that you don't want to blink it in your in your approach to it.

Speaker5: [00:26:58] Hmm. Okay. Sounds great.

Speaker3: [00:27:00] So let's see, you're showing a book here. You're showing the R for data science in the screen. Seems like you have a high affinity for leveraging our for your data science needs. Let me know if you want to add a few a few tips for Abhishek. But before you do, let me acknowledge the presence of newcomers until you for one, Antonio. Great to see you. Joshua and Costa, you guys are here. Eric, as always, always good to see you guys. I'm happy to host here. Don't expect a harpreet level performance, but I'll try my best to make sure we regulate this cool conversation we're having here. Hopefully your Fridays beautiful. And also, if you're watching directly through LinkedIn, feel free to put a question in a chat and I'll make sure to bring it to the panel of experts here. I'm happy to address everything. If I don't know anything, I have plenty of people here are participating will in fact point you in the right direction. With that said [00:28:00] Aunty, welcome again. And if you want to add anything about the data for data science real quick.

Speaker4: [00:28:05] I just need to amend what you said. If Greg doesn't know the answer to something, there's plenty of others here that also don't know the answer to that.

Speaker3: [00:28:16] I'm trying to put you on the spot, Tom, and see.

Speaker2: [00:28:23] Yeah, maybe I'll just add that. One thing at a time. When learning programing language, which doesn't matter if it's oral python. It's. It can be intimidating. I at least remember it being so like two or three years ago when I started, I started with this book and got frustrated and almost stopped completely before actually finding data camp and. That was a new beginning then for me and then other other things after that. But you know. One step at a time. And my journey has been to two and a half years now, and I just started in my first data role this week. So it might take time, but. It has been time well spent, I think also getting ready.

Speaker3: [00:29:34] That's great. That's great. That's great tip right there. A lot of times, you know, we we just we're in such hurry to arrive at a destination. We forget to enjoy the journey. Right. So one step at a time is definitely a sure way of of getting there. And who says, why do we always want to get there? Why don't we just say, let's enjoy the journey until we can no longer enjoy [00:30:00] the what's there?

Speaker2: [00:30:01] Yeah. Where is there.

Speaker3: [00:30:02] Even? Yeah, where's there? Where's there? Right. So I'm constantly reminded of this movie, right? I mention it because my sons always watch it, which is up every time this the guy gets. He wants to get up there, right. Does this adventure. And then he finally gets up there with this balloon in his home, and then he sits down for like 2 seconds. It's like, okay, I'm tired, I'm done. Like, what's next in the adventure? Right? So I love this little analogy of we're going after there and if we actually get there, it's a short lived bliss and we continuously look for the next steps in our lives. So yeah, we. So the original question, Antonio, we're talking about Abhishek was looking into moving into data science and he was looking for some tips in terms of how to go about it. So anything you want to add there in terms of your experience, what has helped you, what angle you you took to approach this? I'm happy to hear. And then Eric has a question in the chat here that I'd love to for you guys to discuss as well. Go ahead, Antonio.

Speaker2: [00:31:14] Yeah. I mean, I know where it's coming from because I was in that same role because when I got into data science, I just wanted to do machine learning and that was it. Anything else? I'm like, Nope, I just want to I want to predict this cool stuff. Well, I got into work and the director gives me one of the, like a file or something, and he's like, Hey, can you figure out this for me? And it was like 200 rows or something. Like there were 100 rows and I started doing like machine learning. I spent the whole week because I was determined I was going to be the machine learning person on the. And after a week, I go to him and he's like, Oh, I don't need this. I got it done. I was like, What do you mean? And he's like, Well, I asked you a week. What took you so long? I'm like, Well, I was doing machine learning, you know, labeling the data the other. [00:32:00] And it was basically just looked at me and he's like, You could have manually done this in Excel. Like this was that kind of a task? And to me, that was kind of like, Oh yeah. So it was kind of more of like the lesson I got out of that one was he didn't care if I was doing machine learning. He didn't care if I was doing it in Excel as long as I got the job done and the people care about value.

Speaker2: [00:32:26] What value do you bring to my company? Like they don't care if you manually calculate it on a piece of paper as long as you get the job done on time for that. So since since then, I was kind of like, all right, maybe like if there's machine learning cases, I will work on it, but I need to do whatever it is. An SQL. I think SQL has brought me so much value and I have brought so much value using SQL and with low hanging fruit. In a lot of companies that like the machine learning is it has specific use cases right where you should use it, but just make sure that you're not too eager to just go straight to that because it's just a tool, right? Querying databases, the tool, loading the database. It's a tool just as machine learning is. So I get where you're coming from, though, because it's very shiny, it's very exciting. But once you start doing it and you start cleaning 90% of the time you're cleaning the data, you're like, Crap, I don't want to do machine learning anymore. But yeah, and also like Russell said, if that's what you're passionate for, don't let us display you from it, you know? Go for it. Yeah.

Speaker5: [00:33:33] I'm not looking for like I don't. Beggars can be choosers, right? I'm not looking for a particular. Am I going to do this? Am I going to do that? I am looking for a job in data science. So.

Speaker2: [00:33:49] You know. Well, then I would say learn SQL because it doesn't matter if you're a data engineer, a ML engineer, data analyst, you'll need to find the data using SQL or something similar. So. [00:34:00]

Speaker5: [00:34:01] I've done some SQL. Yeah.

Speaker2: [00:34:03] Good.

Speaker3: [00:34:05] All right. If there are any other comments for you, hopefully you've had some good tips today and that you've at least encountered many people here who are more than willing to give you more tips offline. So feel free to connect with them or continue to follow. I'm pretty sure there's one Dr. Tom Ives on LinkedIn, so he's going to hook you up with everything you need. So, Eric. Thank you. Have a question.

Speaker2: [00:34:35] Yeah. So I've tried to figure out and read job descriptions and things like that at different times to try and understand what the difference is, if there is one, between a data engineer and an analytics engineer. I don't know if it's just a hype in the title or if there's actually a difference or anything. And so I was hoping for an explain it like I'm five answer to that question. What's the difference?

Speaker3: [00:35:03] I can take a quick stab at this based on my vague understanding and I truly love your question because I can get some help in terms of like knowing what, you know, what it is. So before I go there for Abhishek, a comment from Harpreet. Tune in into some old episodes of The Happy Hour as well. There are a lot of great advices that you can get there if you want to know more. So thanks Harpreet. Make sure you continue to enjoy your time off with your family buddy. So data engineering, in my opinion, it's the I guess. The discipline of helping. Creating value of data and making data accessible to downstream consumers. And for that, you have [00:36:00] a lot of work that you have to do from collecting data, maintaining pipelines, making sure that access is controlled with security protocols and things like that, making sure that the performance of these pipelines are in check, etc., etc.. There may be other things there too. Again, this is my understanding of what data engineering does, and it empowers a lot of downstream teams like data scientists, analytics engineers or, you know, business analyst, where they can go there. At the minimum, the data scientists would take a pipeline built by a data engineer and do some minimum cleanup and then pouring their techniques, statistics and everything like that to generate more insights about the data. So without a data engineer, a solid data engineer team in the state of the art company or whatever, who says or pretends or declares that they leverage machine learning? I don't think they can do it without a data engineer.

Speaker3: [00:37:09] Unless the data science team or the core data science team have that capability of doing it themselves. Now, the analysis engineer, the way I understand it, if I, if I, if I think the word analytics, that means, you know, what you're analyzing. So to me, analytics engineer has more of a domain knowledge about something. So it's more of a reduce scope in terms of like, I'm good at this and hey, Zack, Zack is over there. Hey, thank you, Zack. Great to have you looking at us on LinkedIn. So, by the way, I'm explaining this because I read something that Zack Wilson said about data engineers and analytics engineers. So I kind of like thought about it and it makes sense. So Analytics Engineer [00:38:00] has a little bit of more domain knowledge it could take or he or she could take, you know, pipelines from a data engineer and kind of generate value based on like key insights about about a domain. And it also knows a lot about less about coding, but more about, you know, creating fast pace, I guess insights to downstream people like leadership and things like that. I don't know if I'm explaining well if anybody else has a better explanation than I do for analytics engineer and data engineer. Feel free to chip and I'm happy to learn here. So it's a good moment for that.

Speaker2: [00:38:41] Or if you have an example of it, even if it's a fake example.

Speaker3: [00:38:48] Yes, that would help. A fake a fake example would help. Anybody else can take a stab at data engineering, analytics, engineering.

Speaker2: [00:38:59] So Harpreet while anybody is thinking about Harpreet link to this post from Zach Wilson I guess was from earlier this week. So glad I'm glad I'm seeing this. I said analytics engineers sit between data engineering and data analytics. Analytics engineers at Netflix when Zach was there, we're called Spanners because they worked on pipelines, metrics, dashboards and experimentation. They usually work in a vertical business area where they create valuable insights, requires more business acumen and communication skills than data engineering. Hmm. Okay. And then I think we kind of understand data engineers as being that larger, kind of like what exactly what you said. Greg So I guess that kind of like makes me wonder, you know, like, what is it that? What is it that? An analytics engineer. Does that say a an analyst who might build dashboards and do experimentation? Doesn't do. Is it that they do that [00:40:00] they have some kind of access to creating a pipeline of some sort or making an adjustment to the data that comes in from a data lake or something, trying to figure out. Then, I guess if the difference between a data engineer and an analytics engineer is pretty significant, then I wonder what's the difference between an analytics engineer and the person on the other side of them? If they're like, I'm embedded in a vertical in my role, a couple of them, but I can't quite do everything that's on this list. So for me, I kind of see the difference being I have to ask a bi engineer when I need another variable added or something like that, but I wonder if anybody else has other thoughts.

Speaker4: [00:40:40] Well, amidst this discussion of whether we use fake data or synthetic data, I just want to be brutally honest, especially with Abhishek. It probably gave great definitions, but good luck getting 99 other data scientists or ML engineers or data specialists to agree with what he just said. Also have checked the reason you're in good company is no one can even keep up with the explosion of techniques and knowledge in this field. So you've got to choose carefully. And that's why you need to be the master of your own learning plan. And it'll be directed by the needs of your career. But I will tell you this if you'll really focus hard on the concepts, you'll be able to keep up easier and learn new things faster as you go along. That's that's my top advice.

Speaker3: [00:41:41] Yeah. And I can tell two different different companies called them different roles too. Right. I know companies who are using software engineers as the engineers. Right. They're are the ones who are kind of like putting it all together, integrating with like production apps [00:42:00] and things like that. So. Costa I'm happy to hear from you.

Harpreet: [00:42:07] Yeah. I mean, you're spot on. It's going to be so different for every single company, right? Like, it depends on the scale of your business. It depends on the kinds of problems that you're trying to attack as well. Right. For example, at Max Carlson, where I'm at right now. I'm kind of an engineer, kind of a data engineer, kind of a data scientist, all kind of rolled into one. Right. So our verticals are very much your male engineers who, you know, we have to take care of the data pipeline as well as slip into a data scientist role over looking at the modeling and the analytics coming out the other side. Our horizontals are more our infrastructure team that are looking at, you know, managing the larger infrastructure and the in the back end, how we're set up with queue flow, how we're set up with GCP, how we sort of WAC, things like that, right? So what's the horizontal and what's the vertical is going to vary based on the stage of progress that your business is at the size and scale of your business. Now, eventually, parts of this, the more focused you are, the the less custom each task and each analytics piece becomes, the more you can kind of standardize and say, Hey, look, all of the data that we collect is focused on this area was to the data is going to go through a certain set of standard predetermined engineering into a data lake or something like that, right? So that's where your data engineer is kind of switch from being a part of the vertical and focused on a particular business operation because that's been a very each time to a different company where you're not focused on multiple verticals, you're focused on the horizontal as a data engineer, where you're provisioning the data, where you're provisioning what the data looks like to make it easier for experimentation, right? So as you get to scale, more and more things [00:44:00] will switch from as you switch, as you scale and focus, more things will go from being a vertical action to being horizontal action, right? Whereas the wider focus is the more these things belong in a vertical because it's specific to the task being achieved by that business unit.

Harpreet: [00:44:20] So trying to understand where are your skills aligned to the process and where do you fit in with the horizontal? Vertical structure of the team is going to vary from the team, right? So now I know kind of where my skills sit and I know that in some teams I may end up being part of the horizontal where I'm provisioning data and some I will be part of the vertical where I'm actually using that for analysis. Right. So yeah, it's just about being really aware of where your strengths are, where your skills are, and. I think this it's very difficult to say one size fits all. This is a data scientist. This is a data engineer. There's a big question on who's a data scientist who serve ML engineer, who's an ML ops engineer. You know, let's not even start with that. Right. And then a data engineer. These are kind of amorphous terms at this point. And it's really going to depend which team in.

Speaker3: [00:45:20] Yeah. Well said. Custom. And I love analogies. I know Russell dropped some some goodies in there, and then it will go to Antonio. Russell says, think of a professional kitchen or small kitchen where a small kitchen, the sous chef and the head chef may create everything between them. Whereas a large scale Michelin starred Kitchen will have multiple sous chef pastry chefs at and kitchen runners, yet each individual kitchen will have their own style and balance. It's as Tom said earlier, it's still in the early stages and people are still struggling to define what should be what. [00:46:00] So it really depends on the company needs. Antonio.

Speaker2: [00:46:06] Yeah. So I was, I was looking up a little bit on definitions, on different sites, what it says. So it says like a data engineer. It's just a focus on the technology and like optimizing the data warehouse. Where is the analytics engineer? Kind of knows that a little.

Speaker3: [00:46:20] Bit, but he's.

Speaker2: [00:46:21] More he or she is more focused on like more business focus, maybe taking that data from the warehouse and manipulating it and putting it in a format usable for the data analyst. And then the analyst. Data analyst is the person who might know a little bit of the data analytics engineers work, but their more niche is like digging deep into the data and investigating it and kind of using it. But with that being said, is like what you guys was, every company is going to try to put you in as many roles as they can to save money. I was a bi analyst doing all of the data engineering for one of my old teams. You know, I always tell them I'm like, Call me whatever the hell you want as long as whatever pays the most. That's what the title is. If you're going to pay me more as a data engineer, it's my title to that. Like the work is going to be the same. So I kind of like play the strategic whatever pays the most. That's what I want to be called.

Speaker3: [00:47:22] Absolutely. Absolutely. Fully agree. On that note, one thing I wanted to ask one question I have for you guys. Do you guys believe at some point we'll start separating the roles a little bit more where, for example, data engineers will be more isolated instead of versus building a team where you say, we have to have a data engineer in there. Like, where do you see this going over the next ten years, if possible? So.

Harpreet: [00:47:54] No.

Speaker4: [00:47:56] It's going to take some time for the terms to flesh [00:48:00] out and the best practices to arise. Boy, this is a hot topic, isn't it, guys? It's it's a firebrand. But like even the term and I know they exist, but they're they're unicorns. And I would love to be one, but a Fullstack data scientist, I'm like, okay, forgive my language, but damn, that's there's so much to keep up on. I mean, can even one machine handle all of that? Yeah, I can, but it's cool to be able to be that. But really, if the company's thinking, okay, we're going to we're going to do this, yes, we're just going to have a full stack data scientists and like. And you plan to scale that way? I don't think so. I don't know. I'd love to hear if anyone disagrees because I'd love to be able to be the only guy at a company that could do all that. But the volumes just is so big. I'm going to shut up.

Speaker3: [00:49:06] Across the.

Harpreet: [00:49:09] It's funny, I actually don't think we're going to converge on, like, specialist titles that are going to stay there. Like, I just don't see that happening for a long time. Like, I don't see that happening for a long time. Like, I'm talking on the scale of like 5 to 8 years. B I don't see that sustaining for a very long time. Even if we do converge on a particular set of, Hey, this is what a data engineer is, is what a data analysis is, what a data scientist is, even if we actually do as an industry, somehow manage to come to common ground on that. Right. Two things are going to happen. One, you're actually going to see what happens in the Web dev space, right, where you're while you might have specialist backend engineers, dev ops in front end, the people that get the largest pay are [00:50:00] still going to be your full stack. Two people that are going to be able to, you know, will a deal or everything. Right. That's still going to happen. So you're going to get this amorphous nature of skills in there and B, the field is going to move. Something else is going to come up and change. The platforms are going to change entirely.

Harpreet: [00:50:18] I mean, is it likely that we're going to stick with the exact same tooling and setup and processes that we do now? Is the nature of how we approach data are going to be the same as it is in 5 to 8 years time? I highly doubt it. Right. So I don't see much value in kind of trying to converge and define, oh, what is this? As an industry, we need to define this because it's just going to vary so much. And the fact is, right, what is data is so different when you talk to someone in robotics versus when you talk to someone in telecom and you talk to someone in banking and finance. Right. It's just so domain specific. Unlike in the past when you're talking about web development or app development. Right. That's been far less domain specific than what we're doing now because so much of what we're doing is tied into how we access the data. Right. So it's going to be a larger challenge to converge on this one and that convergence isn't going to sustain for very long. It's kind of my gut feeling about that based on nothing at all, by the way, but.

Speaker3: [00:51:30] Cool, cool, cool. Well said was said. I don't know, Antonio, if you wanted to add something that somebody else was starting to talk to before, at the beginning, if you have any thoughts about whether we will start separating the roles, for example, a data engineer may be pulled out of a data science team so they can search serve the bigger group. If you have any questions in the chat directly from LinkedIn. Feel free to put some questions there. [00:52:00] I'll bring it to the panel here. Serge, let's go to you.

Harpreet: [00:52:05] Yeah, I think yeah, the same thing. I agree with Costa. I think it will go the way of web development. I think for for a while now, I think the focus has been and also the larger page paychecks have been going with people with serious like programing skills. And I think over time that will change. As you know, no code and low code tools become more and more used than robust. I think the focus will become more people that can that can also deal with a lot of interpretation of the data, interpretation of the models, and also more advanced techniques involving causal modeling testing. There'll be new roles for that model, machine learning auditor or eye auditor, and I think that will create another kind of rigor in the field that currently is all on the how and not the why. And I think that will become central to a lot of things going on in the future. But whether they're still going to be a need for data engineers is still going to be a need for data analysts. I think those titles might they might change. I don't think the job function itself won't change. It won't cease to exist altogether. I, I probably thought early in my web development career that, okay, like nobody's going to be using, I don't know, like PHP in 20 years and here we are. People still use PDF and and JavaScript is still a thing and there is still front end and back end development. It's just a lot of these roles have been abstracted in different frameworks. So I think the same thing will happen.

Speaker3: [00:53:58] Cool. Cool. And on [00:54:00] that note, serves you you created this cool mapping the other day, probably like a month ago that you published it. I really liked it. And what was the inspiration behind that? Like you had you had so many things come together there and you really showcase how it was an adventure like at sea, like a pirate adventure. And what was the inspiration behind that and how did you come about adding different, you know. I guess roles that people not conventionally thought that would be part of the data science journey from actuaries to operations research with which I have a background in, in other other disciplines.

Harpreet: [00:54:43] Well, I, I had this event to newcomers to the field. I was invited to speak at this developer conference. And it was to people that were interested in data scientists, but they weren't data scientists yet, or maybe weren't even thinking of being data scientist. And I thought, well, how do I introduce them to this topic? In a way they haven't thought of it before, because especially becoming it from a developer point of view, your point of view is, okay, all I need to do is get better at Python and and learn to go model fit and that's it. I'm a data scientist, as I explained earlier, and I wanted to show them there's a lot more to that and there's a lot of different areas. And you can come of it to another angle, which is more about the data. So like a more data centric viewpoint. And I think a lot of these fields you mentioned are inherently a lot of, you know, very data centric, you know, actuarial science and also a lot of business intelligence, very data centric. And people don't see it as necessarily always data science. But it is it is more I think it's closer [00:56:00] to data science than, you know, a lot of ways people think of machine learning, which is just simply from a software engineering standpoint.

Harpreet: [00:56:07] And don't get me wrong, I think software engineering is important, but that is that is more about the how and less about the why. And I think that's at the heart. I think it's a lot like, you know, like if you want to be a really good carpenter, you have to know about wood. You know, you have to be an expert on wood. It's not about knowing how to use all the tools. It's it's knowing what you're chopping and and, you know, what kind of what kind of tools you should use and when you should use them. And I, I worry about people going so being so tool centric because they're not really learning the fundamentals of the field. So that map is about, okay, let's explore all the area and play around with it, you know, and learn how to apply it in your domain and your field before you're like all crazy. Like Antonio said, like, I'm going to do machine learning on my first day and I had an intern a couple of years ago when he started, he was like, Okay, I want to use deep learning. Where can I use deep learning? And I had to like, calm down, we'll get to that. You know, first clean this data, right?

Speaker3: [00:57:16] Awesome. Awesome. Serge, where does an architect fit data architect fit to this? This is a question from Costa.

Harpreet: [00:57:29] I think data architects are essential. I think like people that do the different kind of modeling. I've heard people like people that call themselves data modelers at conferences. They've you know, they've said, well, I'm a modeler, but not the modeler, you would imagine not machine learning model or a data modeler. So they create models. They architect, you know, where data warehouses and and come up with conventions. I think that rigor [00:58:00] is very much necessary in the field. I think we there's nothing more dispiriting for a data scientist than encountering a data swamp to tell you, here's my data lake. And it's really a data swamp. I've had to deal with that. It's it's a nightmare. And you just wish someone had come along and made it more navigable, you know, because it just you spend so much time cleaning that you can't get to the what you you consider real work. And honestly, it is data cleaning is is not what we're intended to do. We're intended to find find value in that. But if we spend all our time cleaning it, it's really hard to find value, whether it's through like insights, dashboards, KPIs or models. We just can't find it in Dirty Swamp. So I think an architect helps kind of build. If they're building something from scratch, they're going to make it a proper data lake. And if they have the unfortunate task of of kind of re-engineering the whole thing in a way where it's not getting swampy over time, yeah, it's very much appreciated from our end of the of the field. But I think that architects are integral to that kind of data science, I guess ecosystem.

Speaker3: [00:59:24] Cool. That's awesome. And what do you guys think about the role that philosophy can play and in data science? I feel like it's helped, you know, software development a lot. But what what what about data science? Do you see it more? Do you think it's needed? And where do you think this is going? You've raised your hand. Be fast.

Speaker4: [00:59:57] And I hope Surge [01:00:00] will back me up on this. Asking if philosophy has a place in data. Science is like asking if the water goes in the pool. This is philosophy. It's a branch of philosophy. Now, how I actually. Boy, I don't like to talk this way very often, but just because the importance of this question, I'm going to go ahead and just say. I am a doctor of philosophy. That's what my PhD means. And I took that very seriously when I was getting it and started realizing this is a branch of philosophy that has to deal with. Best practices in the field of designing mechanical and multi physical things. And then when I realized how important predictive modeling was in all of that. It extended. And then I realized, well, this means that all of STEM is just a big area of philosophy. It's not all of philosophy. But when you look at the principles and the arts of philosophy that it's really a set of best thinking practices, how we get from A to B to understand the world better. That's what we do. It's this is. Integral part of philosophy. Everything that we do, if we don't think that way, we're actually falling short because we're always trying to do things better. And that really to the betterment of mankind, where we're examining the way we.

Speaker3: [01:01:52] Oh.

Speaker4: [01:01:54] So that is a part of philosophy.

Speaker3: [01:01:58] Storm cost [01:02:00] of.

Harpreet: [01:02:02] This is more kind of adding to that question than maybe answering it. But I think, like I've been listening to a lot of reading some of the like classic original kind of texts and books within the software engineering realm. I've been going over that again, right? A lot of Uncle Bob's stuff and some of the stuff out of there. Like I literally started reading the what was the old C book, that kind of guy, and Ritchie I think it was. I started reading that again just to kind of go over some of the what did we think were the basics and the key parts of programing from back then? Because that grounds me and a lot of what I do as a robotics engineer as well as a computer vision engineer. Right. So I'm kind of curious regarding the philosophy of data science. I feel like back then, if you went back to ask people, what's the philosophy of software engineering, it was a question of how do we operate as software engineers to bring value to the world? Whereas now most discussions on the philosophy of data science is less on that. How do we operate? It's more on the what problems are we focusing on, the ethical aspects of it. So I'm wondering. What's the space for both of that going forward like? Whenever we talk about philosophy of data science, we're rarely talking about the how of the operations.

Harpreet: [01:03:32] We're also we're typically sort of talking about the ethics and the and the why. Right. But I'm curious to see and I guess the time is the only true answer to that, right, is what's going to happen in this next decade, because we've had that foundational decade of, you know, splurge of technology. And you saw this a software engineer use engineering. You saw that, you know, seventies, essentially late sixties, early seventies, the foundations of software engineering kind of settling into [01:04:00] its scores. And then the philosophies of, okay, there's bad ways in good ways of doing things started coming out and then the whys of software engineering started coming out. You're going to see a similar thing start to formalize itself. I don't know whether we're going to have like a rogue summit of agents from different companies come together for it, or we're going to have too many of those that they never agree with each other. Maybe that's part of the problem, but I'm not sure. I'm really curious to see where that goes. And maybe in ten years time we'll have something like the Agile Manifesto, but the data science, right?

Speaker3: [01:04:38] Yeah. I've been wondering about that too. Will we see an Agile Manifesto version for data science? Erica Sidhu Easier said. Ethics Experimental Study Design, Data Collection, Storage Analysis, reporting. All of these involve considering how to avoid doing harm to the individuals who you're studying or their data and to make sure you're spreading true knowledge, which are findings. I make sure there's a firm foundation of validity to data collection and analysis before you tell people what they mean. So yeah, happy to would love to hear anybody on this to build on cost of thought process here, if you have any. Floyd Gizzi writes it, I will say. So what do you think? You're muted.

Harpreet: [01:05:39] Um. Yeah, I think I agree with Gustav. I think a manifesto would be useful. It's just when a technology is so new, it's kind of hard to come up. I mean, you really have to be a visionary to think, okay, how will this be used? You know, what are the boundaries or constraints we should put in place? You [01:06:00] know, you don't want to hinder innovation, but at the same time, you have to realize what you know, what kind of challenges and potential threats the technology poses. I guess we're we're at a point where it can all already be anticipated, where you can already kind of write something. And yeah, I think it's just the question of of of getting, getting, you know, the visionaries in one room to to write it down or maybe a single one like I nominate Greg Laurie.

Speaker3: [01:06:41] Jcc It takes a village, right? It takes a village to think long term and and figure out what what can be done. I think a lot of things are happening now to kind of mold how things should be, which is we're constantly if we're saying that AI or ML is new or data science is new, how can we as a community or a larger community think about, you know, who to involve, right? So let's not keep it as a closed environment where we're not inviting, let's say, a psychologist or a lawyer in or somebody else or lawmakers or whatever. And to understand how to use this technology while you're protecting, you know, the well-being of society or things like that, right? So it's long term thinking. And also the most complicated thing is how do you do all of this while accounting for a different culture? Right. So culture is something that is very important. You know, based on the country that you're in, you may value something that another culture doesn't. So how do you do all of that with the same system? And that to me is the big challenge. I mean, you want to add anything else?

Speaker4: [01:07:56] Yeah. Oh, sorry. My reactions are going [01:08:00] crazy here. Oh, I meant to start this with Lord Geek or. Gc Lord gee. Cc No, it's geek stuff. It just so you all know, we're, we're teasing Greg in the chat he is today he is Lord Gary Gregoire Kikuyu. And no, we're not mad at you, Gary or Greg. No, but seriously, to the point that we're making right now and how we all wanted to welcome and encourage Abhishek into our brotherhood or excuse me, our sibling hood, because we have some really smart women among us, too, is this think about the most exciting cutting edge area of data science. And by the way, to me it's not Transformers. They're super cool, but I still think the pinnacle is reinforced. Learning now think about, well, yes, there were a lot of petty mathematicians that made that possible, but who was the father of reinforced learning? A Canadian psychologist who decided he really wanted to figure out how to mimic the way organic organisms or organisms learn. And he did it. Now he did it with math. But I think it took that that psychological mind as the foundation rather than the mathematical mind. He did learn the math, but it was Richard Sutton. And that's why we really need to welcome and encourage these people from other fields coming into our field because they add so many new concepts to what we're doing, so much extra thinking. Well, Pavlov was important to, of course, and so was I will fail [01:10:00] to name everyone but. Markov and Bellman. All these important people during the force learning going off.

Speaker3: [01:10:10] Awesome. Awesome. I love I love that that anecdote about reinforced learning. So thank you, Tom, for this for this insight as cost of you want to add some more stuff to hear those.

Harpreet: [01:10:24] Yeah. I guess the the distinction between the time at which the Agile Manifesto kind of came together and now is back then they were comfortably operating in the shadows. They were able to meet as experts come up with an opinion, really boil it down to something and then slowly disseminate it through the work space, right? But now, because of the sheer density of communication that we're seeing, the availability of communication that we're seeing, I mean, the fact that we're able to have this discussion right now from across the world. I'm sitting in Australia on a Saturday morning and we've got people from all over the world, from the UK, from the US, Canada, you know, you name it, right? That changes the way in which we operate with that. So like we're less we were forced to be more open, right? Which inherently also means anyone, including myself, knowing nothing about 99.8% of the like data science field. I can go and Finland. Absolutely. Now does they get the Finns right? Best Formula One drivers ever just saying I'm on another Formula One weekend, so here we go. But essentially it's going to be difficult to really try and separate the wheat from the chaff here. Right. I could go not knowing anything about the data science field right now, knowing less than 2% of it. I could go write a manifesto of my own and I can almost guarantee, if you'd like, market it [01:12:00] the right way.

Harpreet: [01:12:00] You can get a solid couple of thousand people really backing it and swearing by it. And you're probably going to have 400 variants of this. You see it all the time, right? This medium article is like, Oh, this is the right way to work in this. And then another medium article that says Bang opposite suggestion, right? And then half a dozen Reddit threads about, you know, this, that and the other. Like you see it all the time, right? So as an industry, it's kind of difficult to try and agree with each other because there are so many voices now. Right. So how much is it a matter of as a company, we establish what is our manifesto for working well? And what I've noticed is across some different fields, not just in data science, what is working well in operational excellence? I mean, to one company, totally different to another company, right? It depends on the scale. It depends on the personalities. It depends on the nature of the company, the behaviors of the company, the culture. Right. So while I'm curious to see if we do as an industry come up with a manifesto for data science, I'm also like, is it less that and more like a series of, let's say. Sex within the larger religion of data science with our little tome that adds to the collection of biblical content, if you will.

Speaker3: [01:13:27] Awesome, awesome, awesome. And if anybody else has any comments about the power of philosophy in the world of data science, feel free to chime in. If anyone on the chat directly on LinkedIn has any questions, please submit them here. I'll send through the panel here. Since we're approaching the end, he has a list of questions. Maybe we can use that to close it out. And there are a set of three questions. Maybe we can make a quick round from each of you guys [01:14:00] to give a quick powwow here. So he has a sweet series of questions. So in terms of like roll your first role in data science, what was your first do you remember? What was the biggest lesson? And three, what would you have like to know before starting that you didn't? I can take a quick stab at it and then whoever wants to go next. And I think for me, I've always had like a role where I had to tinker with data. So being an industrial engineer, I think supply chain or process engineering was my first actually in the manufacturing sites. So statistical process controls really was my first first, I guess, adventure with data. One of the biggest lessons is never tinker with data to fit your agenda. And one of the things that I've I wish I learned before is always approach a situation with a critical thinking mindset, right? You're not here to be the Debbie Downer or something like that, but always approach it where you exploring your questioning, your forming hypothesis and then go into the data to confirm these hypotheses. Who else wants to go next with this series of three questions? I can nominate to my hosting power invested in me. I choose search.

Harpreet: [01:15:40] Closing remarks. I can't think of anything. What? What do you want to say? What?

Speaker3: [01:15:47] So what was your first what was your first role with Data? What was the biggest lesson you've learned and what do you wish you've learned? You've known before. Before venturing into that role, you didn't know before. [01:16:00]

Harpreet: [01:16:00] Okay. Okay. First role in Data. It's hard to tell. I mean, the first time I used SQL was in my first web development role. I guess that could be a you know, because there was a lot of data I was for online betting site and that was in 1999, a long time ago. So I call that my first role in data because I learned a whole lot about databases, about managing data, about querying data and so forth. What what do I wish I learned? I wish I would have learned Python. Earlier. I wish I would have learned data wrangling earlier. I think I was toiling around with Excel for too long. And what was what was the third question? I'm so sorry.

Speaker3: [01:16:51] The first question is, what is the biggest lesson you've learned based on that first experience?

Harpreet: [01:16:58] Oh, yeah, that first experience. Biggest lesson was. I guess. Well, back in those days, a big deal was how do you keep how keep things small, keep the database small. So I learned a lot of techniques to not only make my code very efficient, but also the database structure is very efficient. I guess that's like a lost art these days, but it was very useful to so I could understand how to make be a more efficient coder in general and to always have that in mind, you know.

Speaker3: [01:17:41] Awesome. Thank you so much. And who else wants to go take a stab before I exercise my hosting powers onto you panelists? Costume. You want to give it a try?

Harpreet: [01:17:58] Yeah. I guess I'd [01:18:00] almost argue. I've never worked in a data role and I've almost always worked in a data role. I mean, I'm a robotics engineer and for me, machine learning was that additional layer of bringing sentience into robotics, right? That's that's kind of my long term goal is how do you make robots see the world and understand semantically the world around them and act in that way? Right. So my like I've always worked with computer vision problems. That's where my focus. So whenever someone says You must learn SQL, I'm like, I don't know any SQL. I'm guessing SQL here. Like, I'm not a sequel ninja by any measure, right? I guess. Yeah. So for me, my first entry into it was more from like a computer vision. It was like a defect detection kind of role, right? So that's where I was working. I was looking at a manufacturing line. How can we reduce defects coming out of it? And it was very computer vision focused, very autonomy and robotics focused. And a lot of what I do continues to remain in a similar kind of similar kind of vein. Right. Is there something I wish I could have learned earlier? I would kind of say python testing. Strategies, right? How do you formulate test strategies for a code base? That to me is really important and I'm still learning that now and I just feel like, you know, even from something like, Oh, you can simulate GCS in your test, right? That's like mind blowing to me. And I, by the way, I found out about that yesterday at about 345 in the afternoon and I'm like, Wow, wait, you can do that. So I'm like, I think testing and and how do you plan out testing on a. On a data focused [01:20:00] system. I'm still learning that now and very much keen to learn more about it. Wish I'd known that a couple of years ago. Was that the two questions?

Speaker3: [01:20:09] What was the biggest lesson? I think you put that in there, right? What was your biggest lesson based on your first experience with data?

Harpreet: [01:20:17] But I think the biggest lesson is actually as a group, whatever whatever space you're working on the data on is understanding why right now, that first role I was in a manufacturing team doing computer vision R&D. Right. I was the only person focused on that. The team's focus was very much split because as a manufacturing team, you care about the yield percentages as an R&D thing. You're way down like, you know, technology readiness level zero one and two. It's a totally different mindset, right? And that can cause frustration and friction and just slow down the process, right? So if you can align your your focus and your way with the teams, why? Right. Everyone who's supporting you and forming that foundation around the work you're doing, if you're going to align on why that's happening, that purpose will drive the the facilitation and the quality of what comes out right. If you can't if you can't come together as a team on why it's being done, it's not going to get done well. So.

Speaker3: [01:21:27] Awesome. Awesome. Tom. Dr. Tom, you want to give it a push?

Speaker4: [01:21:34] Yeah. This one's a good question to me. I really wish I could. Well, there are young people that I counsel all over the world, and it's interesting, Greg. What they need the most help on is not necessarily the data science. It's the hey, it's counterproductive to beat yourself up. Hey, forget [01:22:00] imposter syndrome. You are. We all are where we want to be. We need to just focus on learning and growing. Someone harsh says something to you. So what? Filter it. Take what's true about it and add it to your list of things to improve on. You're in a room and you're not the smartest person in the room. Welcome to the club. Take note of how you want to grow and put it on your list. Just focus on learning and growing and enjoy each step. Throw out imposter syndrome. Throw out. Judging yourself. Throw out, beating yourself up. Just enjoy the growth. You'll get much further that way. Are you in a crappy role where they treat you crappy? Recognize it and keep doing your best work until you can get out of it. Are you in a great role? Don't stop looking for a better one. Just keep learning and growing. Keep pushing yourself. But don't beat yourself up. Don't burn yourself out. Don't ruin your health over any of it. Walk at a good pace mentally. Growth.

Speaker2: [01:23:10] Growth.

Speaker4: [01:23:12] I just I have to preach that to so many young people. It boggles my mind. But then I think. Tom. You would have to preach that more to yourself if you went back in time than to anyone else.

Speaker3: [01:23:26] So I'm grateful for Todd's here. Great closing statements to. Unless someone wants to add something until you want to add something. Or did we answer a questions that you were wanting to to have some insights on?

Speaker2: [01:23:41] Yeah, I know it was. It's a broad question, so all answers were great. So thank you for those. Yeah. Also, I'm not that young anymore, but still, Thomas advises. Always.

Speaker3: [01:23:58] I've heard the message [01:24:00] with all my fibers, right? So it resonates a lot. I've been in different situations and I've learned from a lot of you guys from your insights. So it was very good. Thank you all for giving me some in the audience. Some good things to go go with hopefully is going to be an even better week next week. So in the meantime, please do enjoy your weekend. So let's go ahead and close it for now. And to paraphrase Harpreet, our original host, you've got one life on this earth. Why not go out and do and try and do something super big? So you guys enjoy. Thank you again for your time and I'll see you soon. Hopefully you guys enjoy the conversation. Thank you.

Speaker4: [01:24:48] Thanks. Lord Gary Gregoire, Kikuyu.

Harpreet: [01:24:52] Talk to you.

Speaker3: [01:24:52] Soon, guys.