HH48-03-09-2021_mixdown.mp3-from OneDrive

Antonio: [00:00:09] What's going on?

Speaker2: [00:00:11] Hello, everyone. Hi.

Antonio: [00:00:12] Oh, sorry to disappoint that our pride is not here today.

Speaker2: [00:00:18] Well, all right.

Speaker3: [00:00:21] Oh, God, I've been absent for four weeks, so he gets a one free pass.

Antonio: [00:00:27] Yes, yes. He said he misses as the sun rose against

Speaker4: [00:00:32] Tonio and Harpreet Sahota. You're wearing a an interesting and bright shirt so you could start.

Antonio: [00:00:39] I had to. I had to I had to fit in. I know I can't be wearing this shirt. He's always wearing the nice the lion shirts and stuff. And NAMA or Data, though I had to do the

Speaker5: [00:00:52] The

Antonio: [00:00:53] Rubber ducky for debugging.

Speaker2: [00:00:55] Yeah. Yeah. My.

Antonio: [00:01:00] I. You know, I thought that that was a joke when I first was starting in the field. And one of my professors like, no, it's a then I was like, what do you mean? And the thing is like, yeah, when you don't know your

Speaker5: [00:01:11] Code, just talk

Antonio: [00:01:13] To the ducky and it will just make all your code issues go away or a lot of them. And I've had it ever since. It's a good test. Does anybody else have the ducky or something else they talk to their code?

Speaker3: [00:01:32] Yeah, I'm a dinosaur like nightlight that I have on my desk downstairs that I

Speaker2: [00:01:37] Talk to or read reports a lot. I have this, OK? I mean, it's kind of of my headphones on

Speaker4: [00:01:46] When I'm not wearing them. So it's kind of, you know, if I want to talk to, you know, take up some Shakespeare, maybe a little bit I

Speaker2: [00:01:54] Thing I always used to say I always have a legal pad, actually, when I'm coding. So [00:02:00] because I always giat I can't tell you how invaluable this is because the amount of things you'll forget, the little details, there's there's quite a

Speaker5: [00:02:07] Few, actually.

Antonio: [00:02:08] So now that's awesome. So what's your do you just do pen and paper or do you do like on on your iPhone or something that that counts?

Speaker2: [00:02:18] Pen and paper is fine. But whatever whatever floats your boat is a way that I think it's kind of like rubber ducky, but more interactive in the sense you can draw stuff out. And I guess especially when you're doing analysis, like making sure you're just like writing down. No, like exactly what can has. Yeah. Yeah. I find that it's super useful.

Antonio: [00:02:35] So I like I just noticed that when I have I have things to write down and then or like I have things in my mind. And as soon as I start running down, I'm like, wait, that doesn't make sense. Maybe like I don't have as many things on my mind and as I thought I did or more.

Speaker2: [00:02:51] And you might just be overwhelmed by starting this thing down. I do. I really want to go down this path.

Speaker3: [00:02:59] I have so many whiteboards just sitting around

Speaker5: [00:03:01] Because I go

Speaker3: [00:03:03] Through as a scribbling constantly. If I was using a notepad, I'd go through three a week. So I'm like, AIs, scribble, erase that scrapbook. Erase that. Highly recommend the like eight by 11 whiteboards with the with the pass.

Antonio: [00:03:20] People that are currently joining, they're like, wait, that's not her, something's something's off here. We have a warm Wal-Mart greeter who's joining us. You have was connecting, Eric.

Speaker5: [00:03:34] You're welcome. Hey, what's up? Hello. Well, we got

Antonio: [00:03:37] Another one, you see.

Speaker5: [00:03:41] There are multiple

Antonio: [00:03:42] Erich's

Speaker2: [00:03:42] Now like right now. But still I feel really welcome.

Antonio: [00:03:46] Not good.

Speaker2: [00:03:47] All right. I'm. Now, now.

Antonio: [00:03:52] All right. Yeah, or the artists of data science are preve could be here with us, but unfortunately he's being [00:04:00] tortured at like Oceanside and BMB with a lot of free food and drinks. You'd rather be talking Data science, but unfortunately he couldn't. So that's why. Yes, it's a very, very llave. So I thought I'd try to amoud. I even got myself a nerd for this occasion. My wife said I look very weird, but it's OK. She doesn't understand much fashion as it does. So just before we start, we want to make a quick note that this week on the artists of Data Sciences, a new episode with Tiffany Shlain, if only Shlain is a founder of the Webby Awards. And this the conversation is about thinking one day a week off from technology to really focus, focus on yourself and, you know, just leave and just unplug. So what is start this week about something not related, I guess, to two Data science? Thank you, Russell.

Speaker5: [00:05:18] You can go and

Antonio: [00:05:19] Have the Fisher game. No problem, Russell. I'll give you some advice afterwards. But I wanted to ask something that was on my mind. I know a lot of you guys here. Well, we're all somehow related into Data. But I'm going to open it up and ask, did you think that you were going to in this field that you were in, rather? I know it's like a little bit weird because Data is kind of a new field. It hasn't been around forever. But what was something that you thought you were going to do growing up or maybe when you were in school, college compared to what you're doing now? I'll open it up and ask Mark, did you always [00:06:00] wanted to be a scientist or a Wal-Mart greeter or anything else?

Speaker2: [00:06:05] Data science came out of left field for my for my career. Ais One hundred percent think I won't be a doctor. So I was even in my master's in community health and prevention research at the School of Medicine when I had to like make me more competitive to go to med school. And when I got to grad school, I was like, whoa, this is not for me. I don't want to be a doctor. Was probably one of the best decisions of my life because it is like a like eight years in my life of no longer being in school and training. But, you know, at the same time, I still learn like stats and ah, and that's game assassins' to declare transition with my skill set from from grad school was dance. And I was in Silicon Valley. So just being around tech as well. And so I just jumped into it and became super passionate about it. So Data science doing kind of part of my life for like four years. Now that I say it out loud, it seems like actually a little bit a time.

Speaker5: [00:07:01] Right.

Speaker2: [00:07:02] You know, I was much longer thinking I was going to be doing some health care medicine thing for a very long time. So a lot of my friends are in med school and residency now. And I'm the I'm the lone wolf. I'm not necessarily lone wolf. All my friends, I am. But in the end here, it's all Data community. So but I'm the different runnable among my friends. So Data stuff.

Antonio: [00:07:28] This I know it may be getting a little personal, so you don't have to answer, but there were some something I know there's a lot of people call who are trying to get into it. Was there anything like I know it's always comes to the parents where you told your parents, I don't want to be a doctor and I'm going to be a data scientist. Was there any conversation there?

Speaker2: [00:07:46] Actually, you know, my my parents are pretty pretty chill like my career path was. If anything, I think they were happier. And the reason being is like they saw how much I was grinding myself to kind of become a doctor. [00:08:00] Like the level of commitment I was making an unhealthy way. So they were more so relieved. You're like, oh, my God, thank thank you. You're leaving this like you're kind of running yourself into a wall. So they're actually happy that I left that that path.

Antonio: [00:08:14] Marcel Mian, thanks for thanks for sharing, and I think a lot of people could resonate with that. You can. Did you always think you were going to be a you were.

Speaker3: [00:08:24] Know, it's funny you say that that market

Speaker5: [00:08:27] Versus both my parents are

Speaker3: [00:08:28] Doctors and both both my grandfathers were doctors and both my great grandfathers are doctors. So there is a little bit of pressure. And I'm an only child for some perspective. There's a little bit of pressure, a little bit of expectation that I would go into the field of medicine by my parents. And I obviously did not go that route. I had no interest in going that route. But it's refreshing to hear that other people's parents were not as or like excited about the prospect of you not going into medicine, actually. It did turn out well. My kids are very happy with what I do now.

Speaker5: [00:09:05] But there are some interesting conversations.

Speaker3: [00:09:06] And until I was like in my late 20s, my dad is like, it's not too late to go back to dental school. You could always just like make that pivot if you really want to. And then when I started, like, really enjoying my work and making some decent money, they're like, oh, like, oh, what you did is really cool. What are you talking about? We never stop that.

Speaker5: [00:09:23] I always thought I was going to be

Speaker3: [00:09:25] A professional athlete. That's all I wanted to do

Speaker5: [00:09:27] As a kid. I wanted to control baseball.

Speaker3: [00:09:31] And after I got injured, I wanted to play professional

Speaker5: [00:09:34] Golf and I

Speaker3: [00:09:35] Tried. I went down and I tried to play professional golf for about six months. And that was after I did start learning some economics. And then the more quantitative disciplines, Buckley realized that that pursuit of of that lifestyle and that experience was not viable from a pure Raef perspective, [00:10:00] you know, Dsdj.co/artists was not good enough, but it was it was a good experience to help me realize that I still loved sport and I loved combining Data with sports. And that's what eventually led me into what I do. Now, working in sports analytics is I wanted to still be involved. I saw that I had a fairly unique combination of skills to create value in this. And I just kept pulling on that thread and asking questions that I wanted to know about sports mainly for my own failures. So, you know, I wanted to know why I wasn't performing well enough, why I was not able to compete at the highest level. And Data was the was the way that you go about answering those questions and you go about understanding that playing field. So I was way off in high school. If you asked me if I would ever write a line of code in my life, I'd have been like, you got to be joking me. But but it ended up here in some weird in some weird, obtuse way, and I couldn't be happier.

Antonio: [00:10:56] Nice man, that's that's awesome. Yeah, I think a lot of us then, I mean, I know

Speaker5: [00:11:01] Girl, I thought I was going to be

Antonio: [00:11:02] President, but then I moved to America and I realized I'm never going to be eligible. So the only way I have to go back to Macedonia. But I didn't even know analytics was I came from like Macedonia. And my professor would be like, oh, there's something called analytics. All like, I don't know what you're talking about, you know, but I figured I'll try it. And he he gave me the Titanic data set and told this story before once on the office hours. And it told me that I was like going to die on the net. I was like, holy crap, this thing is cool and know so, so much. And then, you know, you just give it a chance and you take a pass and you see you see where it takes us sometimes. But Kevin, you also say you were a featured developer and project manager, do you want to do better about that?

Speaker2: [00:11:55] Sure. So I did my undergraduate degree was an industrial age that I was a teacher [00:12:00] and I taught teach and science and industrial arts and computers, actually. Then I got started in AutoCAD and teaching AutoCAD, and then I got a job doing AutoCAD. And then my boss knew I knew how to develop, but he knew how to talk and know how to talk plain English to people. So he pulled me out of being

Speaker5: [00:12:23] A developer

Speaker2: [00:12:24] And being a B.A. and going to get requirements to talk into the engineering textbooks and finding out. Because what he heard me do was I'd talk to engineering tech. Then I go to the to developers and I pull up the code and go to the particular line and said, you need to change that. And so he knew I knew how to get into it. Anyway, so then I've been the project manager for twenty five years. And then I started reading about Data science and Data and Data, Data, Data, everything coming across my email subscriptions just started talking about Data. So I've been having fun. I've been playing in Python. I took the Google Analytics course. I didn't do the capstone project. I probably sat down to do it so I could finish that course up.

Antonio: [00:13:05] Yeah, Albert Brayley Drill's people who really do the course, but they don't they don't do their capstone. So if he's watching this, maybe I think he's going to be on your case afterwards.

Speaker2: [00:13:16] Yeah, it's just, you know, you know, it's like go through the whole course, right? You're doing this little Data. This little Data is it's like, that's easy. That's good. Then you get this capstone project and these humongous datasets, like, all right, little overwhelming at first, but I don't know, I might go back and do it. But so I've just been, you know, doing that and. Doing the LinkedIn learning, doing python courses, the sequel classes and algebra.

Antonio: [00:13:47] Do you think that course should have been like challenge more from the beginning? I know Googling Alba would like some feedback.

Speaker2: [00:13:56] No, I mean, I think, you know, it's been good to learn. [00:14:00] I mean, you know, I was a single DBA for eight years, and so it kind of brought back to Data where I used to be. And now so it's just been interesting. But I never really planned on being in data science, just started reading it. And I kind of said, you know, if I want to be ready for my future, I think I'd better start messing with Data.

Antonio: [00:14:20] So I like that. No thanks. Thanks for sharing that. You know, so can I think this is a little bit arrogant to what you do with your 66 days of Data. What do you recommend to people? You know, sixty six days. It's building habits slowly. You know, you want to be committed. What do you recommend to people after dusk six eight? I know I was trying to do something around no code. And then afterwards I finished like my my thing like it was like a hundred days. Well, I actually you're not going to throw. But then afterwards I'm like, OK, I don't know

Speaker5: [00:14:54] What to do anymore. What do you tell

Antonio: [00:14:56] People who listen to you like that or like Kevin, who's like, OK, where do I go now?

Speaker3: [00:15:01] So I honestly believe that 66 days is just part I mean, that's that the whole point is,

Speaker5: [00:15:07] According

Speaker3: [00:15:08] To James Kubernetes book, Atomic Habits. Sixty six days is the average amount of time that it takes for someone to ingrain a new habit. And so you could tell a lot of people continue beyond the 66 days. There are people who are almost at a full year now. But I think that that's the whole idea is that what you're creating is this habit of daily learning. You don't necessarily have to continue to document it if you don't want to.

Speaker5: [00:15:30] But this

Speaker3: [00:15:31] Is to kickstart you to

Speaker5: [00:15:34] Kind of, again, pull

Speaker3: [00:15:36] It, pulling on that thread and go down this this path of Data time, because at least from AIs, it has been just a constant learning

Speaker5: [00:15:43] Journey.

Speaker3: [00:15:44] It's not something that I feel like I can turn off for more than a couple of days at a time. And so I use that as just a way to reinvigorate myself every couple of months. It's like let's let's reengage in this habit, even if I'm doing this well and continuously, I try to do the 66 days [00:16:00] about three times a year. And for me, that's that's just a way to keep that refresher or to

Speaker5: [00:16:06] Have other

Speaker3: [00:16:06] People continue to hold me accountable and to continue to build systems where people can ask questions and and I can ask questions and get feedback and these types of things. So, again, I would look at that as the start of the journey. And this is, you know, doing the certificate of these things. These are arming you with the tools to do the next thing that you want to do and build the next really cool thing and and get started with that project that that we've all been putting off. I think it's just, you know, it's just fun. It's like there's this journey has no destination. The more we frame it

Speaker5: [00:16:41] Like

Speaker3: [00:16:42] Like this kind of wandering, progressive and exciting path, the better for all of us.

Antonio: [00:16:48] Yeah, man, it's great content in Mars. Six days of court as the best non LinkedIn. Not a competition, but it is some some great stuff. So just as you guys know, if anybody has any questions, feel free to drop in and chat. And we'll ask this is what this is about. I know Mark usually has some question that's lurking his mind. He said, can't wait to talk to all of you guys. I'm sure he asks something. But what about you, Joe? I know you not only went to become a data scientist, but you're actually a recovering scientist. Do you want to tell us a little bit about how you ended up as a recovering data science then? Did you think that's what you were in what you're doing now?

Speaker2: [00:17:35] I mean, at some point or another, I've always worked in Data. I mean, since a long time ago. So I guess it was the stuff I had been doing. I guess you'd now call Data Science. Back then, there wasn't really a term for it. I guess it would be like predictive analytics or statistics. I mean, ever just before, I think you had the additional power that you have now to do, like teen learning and [00:18:00] like a real capacity. I always had an interest in it. But Peter's back then. Get very far on them. So I'd say, you know, we're under TGP as we're becoming more accessible from early 2010, early 2000s. I kind of saw that there was. Maybe a possibility that machine finally done it like a real farm. So, you know, dove into that interest. And I think I didn't look back until I started at. Is back in 2012. It's a really early attempt, I think, automated burning. It was it was a lot harder than it is now, for sure. But what I realized is that, you know, I started noticing, well, OK, so big machine learning. I think it's it's cool. But to do it in like a production setting, especially in an automobile setting, we try to repeatedly build it could take a Data set and produce answers like the hardest parts of the engineering, actually.

Speaker2: [00:18:59] Like, I feel like the algorithms part was relatively simple by comparison, because this is not the end of the day. If you have a good understanding about and can figure that out, I think that that was actually easy of everything. And engineering that feature, engineering in particular for, you know, etc Data, such as insane. So to automate that process, I felt like was way, way, way harder than what we were doing. You know, from that point that got me interested in how how do you build systems, you know, over the years, kept seeing this repeatedly over and over where, you know, especially on the 10 to started noticing more companies are hiring data scientists on base like it was. Data science is not cool. It was no, it's cool. It's actually like this like job you'd get like cool an MBA or something, you know, so that you were kind of stuck in like the back office doing. Data work, it was not a profession. Now it's like Wakool

Antonio: [00:19:52] Sixteen's job of the 21st century.

Speaker2: [00:19:54] Right. Yeah, well, we'll get to that right now. So. So, [00:20:00] yeah, I mean, it is a sexy job, you see, like getting into it, and then all of a sudden we kept seeing over and over and we still see this data science teams get hired and then they can't do anything right. They don't have any data about how these systems. And so inevitably these teams either have to do the data and doing work themselves or they have to have data engineers. And so really data engineering, regardless of sort of a prerequisite to getting data for analytics, data science. Now, I can make an argument that even a proper meeting is also a prerequisite, depending on the type of maturity of your company. So. So the recovering to science is better is more tests, I think. And I just like direct experiences as well as those of my peers who come a pretty early. And I guess they we're doing it before Data actually a term. So that's awesome. So, yeah, I didn't think I'd be doing what I'm doing, specifically the engineer. But, you know, if you're able to predict your career for 20 years to today, I don't think you're trying hard enough either.

Antonio: [00:20:56] I like Balón. I like Gowland.

Speaker5: [00:20:58] That's the lesson

Antonio: [00:20:59] Of the day, though. Let's see. Thank you, everybody, for sharing that. We have some questions here. Mark has a question Russell wants to talk about, and it fees again. And then Eric has a question as well. So, Mark, I know I must messaged first, but do you want to start with any of these or do you want to go with your question or you said you're also one of the world NLP.

Speaker2: [00:21:23] So I'll go. My question frisk is a type of response, I think, pretty well. So, you know, I'm in startups and something that's been top of mind is, you know, when you're a startup and this product market fit and hypergrowth, you're constantly putting out new features to see what kind of sticks and its important features. But it's kind of like you're still figuring things out and trying to collect data at the same time, you're generating a lot of data. And, you know, you have to have a find a way to process that and actually drive value. And if you're like a male, basically, you know, there's like this dance between product features and infrastructure. And so [00:22:00] as you grow in your Data maturity, like when's the right time to start investing? And kind of more complicated, not complicated, but more. And you're in your Data maturity and structure. And I think it's always going to be like a long thing. And it has to be incremental. But it's a huge challenge to convince a company be like, hey, let's go work on infrastructure. When you use those resources to build like 10 more features, a lot, three more features. You know, I guess, like, how do you balance that goal?

Antonio: [00:22:28] Do you want the customer?

Speaker2: [00:22:31] I think it's a matter the value of Data as well. Can you talk about features and investing in features? But just they're investing in a data scientist make a pretty strong argument that deliverables Data creates or features as well. They may not be publicly facing features, but I hope they have as much value. So it's a question. I mean, I mean, my my simple question is, do you want this to fail? Oh, no. And if you do, then I guess don't invest in is what I'm trying to do. That's one way to do it. And I'll fail at work somewhere else where I can be supported. So, I mean, it sucks as it happens a time. Right. I mean, you know, I talk with people a lot where it's. You know, like one instance is you see Data pipelines that are kind of broken, but you want to invest in fixing them, but then the Data is all messed up because it working Data pipeline like what? And then specting answers out of the data. Scientists like what? What do you what do you want? It's obviously can't work quite so quickly.

Speaker2: [00:23:31] Clarify, I guess. Looks like they're both highly valued. It's more like, which one do you focus on this quarter? Like how do you know? Like right now is a good time to focus. And like we can't focus on both. Well, then minutes. So you're kind of like the limitation is you either focus on building features or building infrastructure. You know, how do you balance that dance? Maybe especially if it's kind of the off season, that's been the time that. You have to do the get the swap [00:24:00] each each quarter. One way to do it. That's not a good answer, because I don't think that is a good approach. So that's the conclusion I came to, too. I'm just calling it like it is. I mean, I will answer given what you described. So. It's I mean it. The old saying, you know, you get like a baby in a month, right? It's not how it happens. So and this is sort of one of those things I feel like things might be not thought through, so.

Antonio: [00:24:31] That's my take. I like Ken's answer in the comments, you saw a nominal fee.

Speaker2: [00:24:40] But how can one fall? That's all right, Dana. Thanks again. Now, of course, I mean, I'm not trying to be I'm not trying to be flippant in my answer either. I'm trying to give you the best answer I can. But it just I've seen this all too often where I think there are these trade offs. Right. And you're trying to make the best tradeoffs. And it's kind of like either I make this or the other one. But in this case, I would have to say, well, the answer kind of has to be both are you're going to have to make some some options or you're not going to. And then, you know, what are you going to do for the next quarter? Are you going to sit there like?

Speaker5: [00:25:14] Doing hacker ranks or

Speaker2: [00:25:15] Something, I mean, it's people worried, so it's matters that much effort has to be put into it. Oh, higher turnout of young people. Are you not allowed to

Antonio: [00:25:28] If Mikiko, do you want to chime in on this?

Speaker6: [00:25:32] Yeah, I think. So I think for one thing, that struggle's unique, like Data science, the all companies, like any like any traditional engineering or particle product company, will have this question of like explorers, like do you like continuing doing what you better or investing down doubling down that area, or do you continue to just go wide? And I feel like the only advice I've ever heard is like, don't just do [00:26:00] one. Like you have to do some combination of both. I know. And like designing Data intensive mangeot, like, please let me know if I am misquoting it, I think at someplace. Martin, I think he says if you really have to put a hard this for if you're thinking like like not 3x growth, but like three magnitudes force of growth when it comes to designing like infrastructure that is like scalable and resilient, I think that's like a specific sort of application or kind of context that he's giving that to. But I think like a lot of times, it's not just this black and white. For example, if you're testing out a feature, you'd have to go like, you know, I was going to say balls, I find more like peak versions of these terms. But you don't have to go, you know, all the way down to like building a fully featured model. Right. What you can do is you could do something, for example.

Speaker6: [00:26:55] We were actually talking about this in our in our our reading group. You do like a PANDOR test, for example. Right. Where it's like you sort of give the options and then you have some kind of like in this case, you would have some kind of new rule. So, for example, if you're trying to cement users as part of nature, you know, you could say like if you are above or below, like certain sort of beans or you can tell them or whatever instead of like going out and just building like a model. So there are ways that you can kind of like the risk investment into creating new features and some kind there ways that you can do risk creating infrastructure. A big part is like hiring people like Trinary Data to let you know what you shouldn't. And I think that's kind of more like the WERSHBA is not all times there are kind of like several decent options, but sometimes there is like some options that are like a really bad idea. So if you're, for example, like a person startup and you're only starting with 10 customers is probably a really bad idea to immediately go to like enterprise, like Google level style sort of tooling and infrastructure. That's probably a really bad idea to go off the bat. [00:28:00] And, you know, my staff, maybe we kind of maybe tried doing some sort of almost close to that, unfortunately.

Speaker6: [00:28:05] But if you try to go for something in between, you're probably going to be better off. Right. So I think a huge part of it is, first off, like understanding that there is never going to be this like black and white. It's going to be this like space of options that you'll have to kind of figure out. I think the second part is understanding, like what are some kind of risk both when we do risk it is expert advice. Another way to risk it is for anything where you're thinking like, OK, I have to build a fully featured model is first testing out with like a dummy solution or even like manually labeling stuff, seeing if people even like we'll take that up. And the third part is just having very kind of a clear understanding of what success looks like all the time, right in analytics, where, you know, you do a dashboard report and you show to this part, what does this tell me? And like, oh, this is really is it really? Because it's like growth is like going up like five percent in some industries. That's the thing. In other industries, that's terrible. So it's all about contextualizing what is like success in those KPIs look like. It's very similar for when you're thinking about these like infrastructure sort of questions and explore for both.

Speaker2: [00:29:10] Both answers are really then just reflecting on my past year at a startup and like seeing all the growth through pad and all the new stuff. And like because we have so much new stuff, I'm like, wait, this all this new stuff we need to do. So, you know, trying to trying to like formulate like what I want to do in this company, like provide potential solutions, like trying to make Nonis I make the argument, but like place it within the costs of the company. Just my discretional in my experience where I'm just like, I don't know enough about this. I haven't been exposed enough about this. Actually, I have a fruitful conversation. So this is really helpful. Well, I want another piece to add to if you're a startup, I mean, there's nothing like this thing personally just kind of goes and does stuff and then tells you that you just did it. So I think a lot of value to that, honestly, in the startup. I think that big company. But at a startup where [00:30:00] you are under the pressure of resource constraints and runway in particular and time, I would have something that my team did that and it was cool and it added value and it didn't break in production. Make sure that's cool. Prototype it out and show what you've done first before you put it in production. Maybe. But I would say that just because it's not international, it doesn't mean you shouldn't take the initiative and show people what you did, because it's one of those things where I notice the Data especially, it's like showing is better than just telling that you show something that you've done and just really prove it out. I mean, it shows a couple of things. One, you got initiatives that is like cool and all that stuff. The other thing is you have a demo that you can show engineers who otherwise are just that kind of think about this, because I have other things to think about, like my code base or, you know, my commute, like all the other stuff that's on people's minds.

Speaker5: [00:30:50] Not very few people

Speaker2: [00:30:51] Care about your problem as much as you are. You really got to take the initiative and just like show them what what you want to happen.

Speaker5: [00:30:58] And thankfully, like.

Speaker2: [00:30:59] Right, making a prototype, it shouldn't be that hard if it's just like honestly like this is a work I'm envisioning and just showing people like that's cool. You know, I've had people do that to me and my teams where they I was like, I don't have time to think about this. But they just did it anyway and showed me that that was pretty darn cool, actually do that. So that might be the other approach. And frankly, at a startup, I think. That's kind of what you got to do. So, you know, if they like it, great. If they don't, then

[00:31:23] I don't know. So.

Speaker2: [00:31:29] Thank you. Now that I've finished in major projects, I've had a wall and a lot of time and so I think brought me here. So this is really helpful.

Antonio: [00:31:38] So drop in the address and Joe will send you a bill after this call. But if that helps, I think also there was a good conversation going on in the chat. You might be able to check out some resources there, around there. If anybody else wants to chime in, if now we go to Eric's question, because I think it's related [00:32:00] to that. Ok, Eric, do you want to ask the bigger

Speaker2: [00:32:06] Question is my first time here. It's so cool to have this audience to pose this to basically I don't know if you guys follow Tyler Folkman, but I work under the guy who works entertaining and his awesome Data. It's all about like researching features, failing fast, like getting quick feedback, you know, production early and stuff like that. So he's like, I've really been on the engineering side most basically all my experience like I did. I did like some deep learning walk, you know, like I have like a personal project that I want to launch soon. But mostly it's like mostly it's been like the Mellops type stuff, like, you know, make your models and get scale, stay all, retrain them, stuff like that. And so I just felt like in my career so far, some of the more exciting part of the data science that really just got me really excited. Like I heard Ben Taylor came and spoke in my research lab when I was in school,

Speaker5: [00:32:55] And he just like caught

Speaker2: [00:32:56] My imagination.

Speaker5: [00:32:58] He gives the same

Speaker2: [00:32:58] Presentation every time. But but yeah, that was what triggered me into like going and doing math and stuff and getting into the space. So so yet one thing that was exciting to me was kind of like the inventive nature of like research. You know, like you have all these unknown problems. Maybe you can use data science to come up with these cool solutions to them. And so so far on the engineering side, like my sort of slice of the lifecycle ideas come pretty much pretty vetted, like maybe not maybe. Maybe the infrastructure itself is a little experimental, but like but I don't feel like I'm necessarily part of like the inventive part of that process. And so actually, I think I'm about Divac, my full time job, like like this is like a risk I'm thinking of taking because I just spoke to the research side, something about like connecting with the job and taking a new like part time. Job as a researcher, like somewhere else, it's because I feel like I'm just I feel like I haven't had the chance to see in the research side of data science. I just I just wonder what can you guys speak to that? Like what's it like on the research side versus engineering? Like what [00:34:00] are the different stresses you have or like the different parts that are fun? Does that make sense?

Antonio: [00:34:05] Sure. Yeah. So you're saying you want to learn more about the lot of engineering. We all Mikiko houseware and raised and then will do whoever can Giam. But let's let's start it off and see where we go.

Speaker6: [00:34:24] Yeah, my name is going to be pretty quick because also I've never really sort of been on the resew so few things that there are like Data scientists roles where they don't quite issue research like they do like analytics and experimentation, all that. And then there's the research for research folks, and there's actually a few on the call. But one thing I would say is we don't need to quit your job to go get research experience. The reason I know this is because they had like they had a couple of Covid and prior to Covid, like Data hacker bombs or whatever that were sirf open public. And there is a couple of them where, for example, when we were looking at like this, like

Speaker5: [00:35:05] The impact of

Speaker6: [00:35:08] Red lining policies on Covid disparities. And this was back when I was at Teladoc, and we were very, very interested in the social determinants of health. And myself and to others, we act as sort of like the data scientists, engineers. And we were partnered with a few endemics from who are teachers or professors at a university. And the outlook of that was was a paper that could be submitted to conferences. So that you say like one thing, like you don't you probably don't need to quit your job to get kind of Alastor research experience hands on where, you know, if it's just involved with like Data and occupation and all that. There are definitely opportunities out there. So but that's once again, speaking as someone who liaison with academics for research [00:36:00] to provide engineering and all that insight.

Speaker2: [00:36:05] I love that insight, I mean, just to clarify, I wasn't thinking about quitting my disl my current hours back, but yeah, I see the distinction between like academic research and maybe like more product focused research is probably because my thinking.

Antonio: [00:36:21] Can you have your hand raised is what

Speaker3: [00:36:23] I was going to first ask for clarification that you just gave. I think that was my friend. I mean, if you're looking into the academic

Speaker5: [00:36:29] Research within like Michigan,

Speaker3: [00:36:32] Within these domains, that's a completely different animal where you're focused on optimizing or creating algorithms rather than applying data science to a specific ban. I personally would try and look at their current organization to see if you could add in in some of the meetings and get exposure to what's upstream of what you're doing. To me, that's like a really logical step, career development wise for you as well, where you can say within a company that I'm at one, like, what is the work impacting? Like, where does this go? How does this work? Probably help you do your job better. But it also, you know, if you did want to move up that pipeline to a different role in that in the company, you'd know exactly what to expect. I also wanted to second what Mikiko said

Speaker5: [00:37:21] About there are

Speaker3: [00:37:22] Plenty of opportunities to do research to to work on papers that do very legitimate things within the space already. I mean, I've talked to fairly recently a quite a few academics that are not doing machine learning. They're either in biology and some of these other things. And if you did want to go more on the academic

Speaker5: [00:37:44] Side, they're looking for people

Speaker3: [00:37:45] To program or to actually do the analysis on the research that they're doing, because that's not within their core

Speaker5: [00:37:51] Competency, core

Speaker3: [00:37:52] Competency skill set.

Speaker5: [00:37:54] So you can create good value and you can help people

Speaker3: [00:37:59] At scale [00:38:00] at nonprofits, or that they don't have as much funding. Or you can also do some contract work or Kaggle or some of these other places where they have public Data that you're doing very legitimate things in terms of research there. So obviously, up to you, if you're not going to quit your job, you want to pick up something else. That's definitely an option. There are also plenty of options within

Speaker5: [00:38:21] Your specific company, within sources,

Speaker3: [00:38:23] Which might be easier to get into rather than applying places and figuring out how to pick that up. Just I got some I got

Speaker2: [00:38:31] Just real briefly, Ken, I'm going to believe that you didn't remember that. I mean, the statement I Mikiko said, so be cautious in the future.

Speaker3: [00:38:42] I'd be trademarked it yet by spending.

Antonio: [00:38:47] Yeah. Sorry, Tom, I go to Muniment, I'm asking

Speaker5: [00:38:50] You, but we'll take you court.

Speaker3: [00:38:53] But I just got off of a three hour call with my lawyer, so I'll probably settle. I don't want to pay any more money.

Speaker6: [00:39:03] Could you, al-Sahhaf? That's the one Sternin trademark, what Mickey Kaus said. I don't think he says it enough.

Antonio: [00:39:10] Yeah, but I think McGuigan needs some royalties on this deal, otherwise I would not do that. So thank you. Thank you for sharing. Besides the AIs to add.

Speaker2: [00:39:24] Eric, awesome question and. It put an apple you in the cost. The best way to research is to church. Now I'm going to give you system from reinforce learning, learning an agent and reinforce learning starts that initial policy is crap. Your initial break into research might be crappy, but Brya learn from it, refine it, try it some more, learn from it, find it. Don't take no for an answer. Let's say you come up with something [00:40:00] great, but you get any journal article to it because you don't have those magic three letters behind your name, which is bullshit, by the way. Well, the PhD to partner with.

Speaker5: [00:40:12] And just say, can you let

Speaker2: [00:40:14] Me be probably second author? You can even read sure. It sounds journal article enough. But don't just go for it and. Then the important to leave a trail of any kind of document, formable posts or blog with blog about your journey or what you're doing and what you're learning, and show strong proof for it. In everything that you post. I mean, just be a social group about all this. Grow your list it over time and be able to break into research. Now, it may not be as easy as if you had a Ph.D., but again, you may have to form some alliances to really get. The attention you want, but breaking into that realm and through looking at a French at the edge of the arm, that is challenging,

Speaker5: [00:41:05] But there's so many

Speaker2: [00:41:08] Areas that are needed. So many I mean. And it's like, does it look upon you? Five minute entrepreneurs. I actually think that might be the best advice you get in this space, but it's earch, that really makes a difference. You will get some traction, and then I would also resist the temptation to be a Data science researcher, Data science researcher. I'll just do it. Do it in your spare time. And if you're doing significant enough stuff, you probably eventually get a job doing it. That's my two cents. Awesome. Yeah, I mean, I think I wasn't really like breaking into it, but that is for sure. But I think the community is super tight. And so like you're saying, like if you do big projects, like usually usually there's [00:42:00] someone with an opening on some team at some point that will help to at least get in in the production space. You're smiling until he agree or not. And that's that's how it seems to me like. Do I even want to do it? Does that make sense? Does it seem to me that like always, the burden my analyst friends have had to carry is like proving that they're worth paying, if that makes sense? Like as an engineer, I can always be repurposed if I and the company proves not to be valuable for some reason. So I just I just know, like in all the hotness of like maybe getting to experiment with new like you have like that that weight over you. Yeah. But like are we going to have some sort of like later? But yeah, I really appreciate what you guys have said. It's more like one last thing. Eric and I started learning about Transformers initially. I thought that it was the end of democratized Data science research until this book came out, Dennis Rodman's book on Transformers. And there was a point at which he pointed out, hey, these guys with a transformer model,

Speaker5: [00:43:03] A belt, I think

Speaker2: [00:43:05] One ten thousand the size jpt three. Bring regular hardware with a good GPU good processor in real time on a specific task. Spanked cheap threes, but on performance, it had to be specifically trained and everything. But after I read that, I realized human ingenuity always wins the research base. And now there are transformers. The train by new techniques by Lyn formers, I can't even remember all the names, but there's there's always some mo better clever way to do this stuff. You just got to

Speaker5: [00:43:47] Keep researching that space.

Antonio: [00:43:51] Yeah, and I think that's some good stuff right there for you, Eric. I if you can do it at your company, I mean, sometimes you have downtime. I know I was [00:44:00] on my when I was starting out, I was doing a lot of business intelligence and reporting I wanted to do like machine learning stuff and trying new ideas. And I said to my manager, you know, when I have some time, if you don't mind, I think there's an opportunity with this Data. I want to try it out to see where it goes, you know. So then on like on Fridays, they allowed me and every Friday it would be like I would just have time to to work on machine learning, see where it goes. And I got to slowly build my my experience that way. So that kind of stuff is there always. I know. Also doing things after work here and there. Scrape some data, try to find some insights. I forgot who mentioned they were posting was the LinkedIn. People really notice it and it gets conversations going in and it kind of into the validation. And I think a third thing, if you're doing trying to do like actual research like the professors do, if you have contact with some professors thought of them, because I thought I wanted to kind of do some of that. And I got to one of my professors like, hey, I want to help you out. And he gave me all the data and he's like, okay, this will do. And then it starts getting peer reviewed, and you have to reiterate and I realized real quickly, I'm like this on LinkedIn, I don't need it in some academic journal.

Speaker5: [00:45:17] Right. It wasn't really

Antonio: [00:45:18] For me and for some people it might be. But I quickly found out that it wasn't for me. So try to find opportunities in your current situation that you can do some of this stuff and see where it goes. But Mark has something that he wants to add.

Speaker2: [00:45:35] I just wanted to kind of add to the Tom's point about how you can find and what you're saying or professor. Currently, I'm working on a project with my mentees to classify REST's newborn mortality within 28 days. And it's like this really novel Data set I found through a research paper. I only have like a few views. And this is gold. And depending on how it turns out, I'm just going to go [00:46:00] to my old professors and pitch it to them and be like, hey, I did this analysis already. Here are the results. If you want to do the right up, you can be. First of all, I don't care. And you can handle the publishing stuff like that. And like we can just talk about the analysis. And there's been guidelines like what word counts for you to be a full author for a researchers. So you can check that out. But like that's another thing you do if you find a novel, Data set, do a cool analysis just to professors to collaborate them. Say thanks, guys. Yeah, I'll let you I'll let you know what doing, I think. We did try it on for like a year. So whether it's internally or somewhere else. But something I've been thinking about, it's been weighing on my mind. So it's just dorsum to hear you guys, I have to say. Thanks, guys.

Antonio: [00:46:43] Hey, there's some project posted on your GitHub, write up an article about it. We'll be happy to peer review it for you. And then you see if you if you really like doing that,

Speaker2: [00:46:56] I will literally tag all of you. Here it comes.

Speaker5: [00:46:59] That's awesome. All right,

Antonio: [00:47:02] Mark, I don't know if your hands raised from before

Speaker5: [00:47:04] Or you just reraise

Antonio: [00:47:05] The.

Speaker2: [00:47:09] But I stayed up, I don't know why I went, didn't go down.

Antonio: [00:47:12] Ok, so thank you, everyone. Hopefully this will help or Eric. We have another question, but also, I know Russell had a question from before. So, Russell, once you go ahead.

Speaker4: [00:47:26] Yeah, thanks, Ansonia. So this is touching back to last week, and we got deep into it. But one of the points I made last week was a conversation we had with Mark previously about using an app to use to verify the validity of data, to authenticated that it could then be used for other purposes without any question about its authenticity. And in the last week, there's been additional news breaking about, and it's being used in the commercial market, which is it's made me think about some conversation [00:48:00] we had with Ken a little while back for the Zadran Can. And I think I posted last week there's been a schoolchild in the UK who has created a lot of small, wild images that have sold for hundreds of thousands cumulatively for this school age child, I think maybe 12, 13, something like that. Quite some quite interesting. And what I've seen in the news recently has been the the board eight. Is anybody seen this? There's been all of these cartoon images of apes that have been selling for quite some large fees. I think Stephen Curry bought one which made headlines for, you know, tens of thousands of dollars. And there's a lot of variance for these. I guess it's almost similar to a trading card game where you've got all these variants of a single image with changes in them, and some are more rare than others. So people chase those and low sell high prices. So that kind of a dynamic and it seems to be really picking up pace. So I'm just interested in the in the opinions of others on the Coolibah, but definitely both Mark and Ken in the first instance.

Antonio: [00:49:13] Are you interested or more about what they think about this new trend or generally

Speaker4: [00:49:18] To see how NFTE how the use of energy has been changing just over the last two months, where it seems to be being rather than the artworks that we saw being sold some six months ago when there was the huge 63 million dollar artwork, which was a like a cumulative piece of thousands of different artwork, she had to zoom in, etc. These are single pieces that say similar to a trading card game. This seems to be really picking up Paten. I expect that there's going to be more of these types of things hit the market soon. But there's you know, you were looking with can, that seems to be a little different. You know, that's you know, it's a it's an alternative virtual [00:50:00] horse racing game established rather than just plotting to collect. That's something that there will be interactions with. So where do you see the immediate future when it's and it's kind of picking up more pace on the one of the collectibles type of market or more on the on the Z front of Michael?

Speaker3: [00:50:18] Well, you know, I was very skeptical about NAFTA is from the collective I that. You know, why do people have collectibles? It's a store of value, it's unique in certain sense. And if we're looking at a global market of collectible A.F. tiers, like their digital goods, you can create or generate millions of things that are similar or like what makes them,

Speaker5: [00:50:44] You know, unique

Speaker3: [00:50:45] In a circumstance. And I think once you start creating individual markets around each one, that's where the value really comes from. Historically, where I saw the most value in NFTE is is for content licensing. So one of the biggest problems with content creation on YouTube, on Twitter or any of these things is there is copyright issues with music. You're listening to things you show on screen, whatever that might be. And Nifty is allow you to label those things, show that you have ownership over them so you can use them. And in a perfect world, you could have a marketplace like this

Speaker5: [00:51:18] Where those things would

Speaker3: [00:51:20] Be relatively affordable, because right now, if I go and I want to buy a picture from Shutterstock, it's like seventy five dollars

Speaker2: [00:51:27] Something.

Speaker5: [00:51:27] It's ridiculous. Right. So if you're

Speaker3: [00:51:29] Outsourcing this to individual people, I think that that's an incredible place for this to have real world utility and usefulness. I think that what I expect to see with the collectibles, the market just inundated

Speaker5: [00:51:44] With new NFTE is

Speaker3: [00:51:45] That anyone and their mother could create credit. They're just going to be like home and saturated. And the original pieces that are out there are going to continue to carry value. But what happens when,

Speaker5: [00:51:57] You know, everyone is seeing seeing the value

Speaker3: [00:51:59] Created by the [00:52:00] user, how much money people are making. And it's just like Krypto, like there's millions of old coins now. Right. And what differentiates them? Some have utility, some have. And that's where I think the NFTY market is going to find itself. And that's why I really like that run in my mind. That was the first place where NFTE

Speaker5: [00:52:23] Had utility

Speaker3: [00:52:24] Outside of saying, hey, just own this. I can also, you know, raise as far as I can breathe this horse to create new nephews. There's a system, an ecosystem around it that continues to drive its value and unique in different ways. And there's some other is now I was looking at one where, you know, you buy an individual character and you can buy like weapons for them. And it's like a little RPG game, which I think is kind of cool. Not as well implemented as I think it could be. But the more that we create systems that that make the individual nephews or goods useful in and of themselves outside of the collectibility aspect, the more these come and stream in, the more applications we're inevitably going to. I actually hadn't heard of the Data use case, and I love that. I think it's inherently practical. It makes sense. There's a lot of Data marketplaces out there that don't do this that well and keep track of where all this data is going to be able to understand who has legitimate, not legitimate use of these things. That could be the next evolution in my mind. It's embarrassing that I hadn't put those two together because I see the value and the utility and both of these things. But yeah, I don't know exactly. I kind of went down some rabbit hole there, but I love that direction. The more we can make these things have practical uses and create ecosystems, the better.

Antonio: [00:53:48] You know, I like to be friends by Gary Vee Spirit has 10000 be friends right there, all each unique. So if you if you own them, you get [00:54:00] access to his conference. Meet Prince Decolonizing is called for the next three years. So what which I think I mean, he could he just sell tickets probably to a conference. Right. I think with the nafees allows him to do it. But if I buy this three year access next year, I don't want to go to the conference. I can just transferred out to somebody else. Right. Maybe I got my value out of it. Now somebody else gets to do that. And I mean, it works for him as he gets a royalty on every tronson. Right? He can't do that. If it was just a regular ticket on like the Grand Master, Ticketmaster will make the money every time. But they could sell him as the creator wouldn't touch any any part of that. So I'd see some of those those uses. I've been thinking about it. So there's this pizzeria, actually, that I go down the street and many royalty. I'm like, I'm here every day. They don't give me anything.

Speaker5: [00:54:53] I wish they elegant NLP.

Antonio: [00:54:54] Right. They say if owner a holder of this nifty gets 50 percent off on on pizza, for example, and then I use that for as long as I live in this area, maybe I buy it for like a thousand dollars or something. And it has like a cool art. And then I use it for long and then move out of this area. I don't want to just lose the thousand dollars. I can just pass it on to somebody else. And then you just sell the INFP and then the next person goes on. I do see a lot of those kind of smart contracts that could be used in the future. I think somebody asked, do we actually need a lot of a little puppy to go with it or like an ape? I don't know if you need the picture. And it just adds on to the flavor and just makes it look interesting. But I do definitely believe and in the smart in the smart contract of things, Marc, do that any day.

Speaker2: [00:55:49] Yeah. I mean, I can can set up tzavela. I think the thing that really excited me wasn't the images or like collectibility, but it's just a new way of thinking of ownership and that new way of being of ownership. [00:56:00] For me, it meant there's new ways to create business models. And so what really struck me was like, what's the level of innovation of thing and how to monetize this? That's going to happen. And so I think that the original post that you're referring to was I was thinking of entities as a way to mitigate the impact of Dietrich's. And so particularly for like government agencies. So say, for instance, is making something out like the president, United States put out a press release. Right. And that's a video. People can easily change that, I would say, whatever they want. And so it could potentially be a way to say, like this entity attached to this exact video, the government owns NLP. And I can say with confidence, this is the actual true image or video and wasn't altered because validated by the ownership. In addition and it's say, for instance, like some bureaucratic process where it goes from the White House to the Library of Congress, they can now pass that on to another department and it's track on the ledger.

Speaker2: [00:56:59] So there's like this transparency component as well. And so it's not necessarily like, oh, yeah, I own these images. But as more so thinking, what is ownership in a digital world look like? And the whole thing with with like block chain is like just me from just talking to Carlos and reading his books. I like to be the best person to to talk to us by essentially that you're removing this third trusted party and you're taking ownership of the trust and processes around validating things. Right. And so by doing that, that makes things really efficient, because now you have to go to the middle person to make these transactions happen. So that's what really excites me, is that there's just a new opportunity for business. And I was just really excited for what innervates said to drive, kind of like like you said, with to be friends from business models around ownership in a digital space.

Speaker4: [00:57:50] Yeah, I think that's one of the more more exciting and easily is that that that ownership for content, as you say, to protect against deep fakes, [00:58:00] etc. I do like Antonia's idea that for kind of, you know, discount loyalty. Jards, that sounds like a great, great use case for it as well. But what we went on to talk about last week, illation a conversation which kind of ties back to earlier in the call this week was so you had a model that had a great training set. You could you could nfte that training. So to make it available to all the users to, you know, get two months off from from, you know, and sort of start from zero percent, start from 30 percent. You know, it's not going to be exactly what they need to go up, but it's going to be a better starting point. So you could nfty that can yield some kind of monetization from it, depending on what your

Speaker2: [00:58:43] Your

Speaker4: [00:58:44] Requirements are. But all of those types of things, I think there's there's great prospects for the future of NAFTA. For proof of provenance and proof of authenticity, validation, et cetera, I think that's where they're going to be stronger. This whole collectibles market, I think, is going to be a bit of a flash in the pan, as Ken was saying, you know, a great bit of digital artwork. Someone can sniff that from a screen printed out. And, you know, no one's going to be any any the wiser unless it's really well broadcast that it's a big piece that some are well known, has ownership, too. And if someone on the other side of the globe copies it and puts it on their wall, you know, on a note, so very typical police, the ownership of a of an electronic item like that, I think.

Antonio: [00:59:34] I think it has a lot of potential, I mean, you can while you have you can have like a hundred fifty thousand subscribers on YouTube, it those people are dying to have a one hour conversation where values can create. And the whoever gets that NFTE, you get like a private one hour call or like us advice session with with Ken and once every year or once every six months. Right. Maybe that person who

Speaker5: [00:59:59] Does limit the call [01:00:00] does

Antonio: [01:00:00] It for two years and year three, they're like, well, Gumby can stop me. So many things now. I'm ready to take

Speaker5: [01:00:07] On the world. I can

Antonio: [01:00:09] Sell this. And every now and they sell it, you know, so they can get a royalty in his values increasing by then. He has two million followers now his. And if he's worth like a hundred dollars. So I think something like that is is bound to happen like creators in word. And there could be a definitely a growth. I do think one of the other uses, though, is making it more accessible to people. I know when I started,

Speaker5: [01:00:36] It was like

Antonio: [01:00:37] You need to have on Coinbase, then I need a metal mask account, like a public wallet, but I need to buy the NLP from open sea. And then there's like gas fees that you have to pay, you know, so it's like kind of like three

Speaker5: [01:00:50] Or Brams who are who you need

Antonio: [01:00:52] To actually just buy the Nifty. But I think from that perspective, once it becomes more accessible to people and anybody can just go buy them, then we'll see where it goes. Where can F.T.? I like that. So we he said if you buy, he has a new book coming out 12 and a half. Is that for every 12 books you buy, he'll drop you like an NFTE as a surprise on November 16th when the book comes out. So I missed out on be friends and I didn't want to regret it and miss out again, and all my life I'm like, I'm buying some books. Like, oh, how many books actually ended up buying? Eight hundred and eight copies of the book on Amazon. And I'm going to get I don't know what exactly is. I think it's nine and it's from Gary on November 16 there around there. So I'll be happy to share one. That is it is like an investment, but I've been. That's why I [01:02:00] did a giveaway the other day. I would've then I'd be happy to send you guys and books as well that are on this call once it comes out, because I have one hundred eight books, and I don't think I need that many to read.

Speaker5: [01:02:14] I didn't know too many.

Antonio: [01:02:15] Like with the board, abes projects were very expensive. So I was like, what the heck? I'm going to take a chance on this. See what happens. It's a learning experience. So I'd be happy to share my my experience once I do receive this.

Speaker3: [01:02:29] Well, I am one of the projects I'm doing in October and November, I am going to be scraping or you just using the public eye of open sea and getting a lot of data on

Speaker5: [01:02:40] The marketplace there, mainly

Speaker3: [01:02:42] For Zadran, because I'm actually like there's cool outcomes associated with that, but I'll be sharing that. So everyone keep an eye on. You want some more like Data insight into what that space is all about?

Antonio: [01:02:57] So you're going to start your sixty six cans of Data with Berkeley. As anybody else would like to add, I don't know. And so we will keep going. We'll keep it going all night. Russell is the only one doing a right with a drink in his hand. I should have grabbed something, but it's all right. Ok, I'll catch over afterwards. Let's see. We had a question. I don't know if he's from Mark. Mark, do you want to go ahead and ask your question?

Speaker2: [01:03:28] Sure. Thanks for listening. So this project I'm starting really just trying to track hair and movement

Speaker5: [01:03:37] Of the camera angles

Speaker2: [01:03:39] Basically off the table and just trying to track and movement around the table

Speaker5: [01:03:44] And

Speaker2: [01:03:44] Around there, like maybe touching a few different areas. And I just want to track they a bounding box like true or false? There's a hand there and is not there. And some of the things that I've been trying, like image differencing

Speaker5: [01:04:01] And [01:04:00]

Speaker2: [01:04:02] Pretty much the contours are created from that. Like I'm not really getting much, especially because of the frame that I have. And so I'm also considering like low power or

Speaker5: [01:04:15] Computer Covid usage, of

Speaker2: [01:04:18] Course,

Speaker5: [01:04:20] Neuro networking.

Speaker2: [01:04:21] And that's where I was just steer away trying to do like really simplified. But I don't know if that's the best way

Speaker5: [01:04:27] I am think,

Speaker2: [01:04:28] Because based on my research and reading, the going of a big stack overflow sources. Seems like a combination between like the broken CD packages on

Speaker5: [01:04:40] And

Speaker2: [01:04:40] Price and your networks, like NPIs, like Baie Points is really probably a good avenue. But they know. Trying to like going something

Speaker5: [01:04:51] That's trying to basically gather,

Speaker2: [01:04:53] You know, or suggestions to basically tackle such a problem.

Antonio: [01:04:58] Do we have any expert who would like to chime in on the Syrians? One who's put on the question, you're saying how to get stop and stop.

Speaker2: [01:05:14] It pretty much. It's basically, you know, recognizing and I see a lot of tracking like this, Chayefsky in a sense

Speaker5: [01:05:26] Of what

Speaker2: [01:05:27] It's like in a regional interest in a contract, stands for few friends. But because of a to frame the second type of frame rate, it loses a hand and. Basically isn't detected again, so I

Speaker5: [01:05:41] Basically have to

Speaker2: [01:05:42] Reapply and there.

Antonio: [01:05:44] So are you using any tutorial? I know Adrian. I can't forget his. I forget his last name by is Ohmes. I think it's called Image Search. Are you familiar with that?

Speaker2: [01:06:00] Heard [01:06:00] of it?

Antonio: [01:06:01] Ok, so I don't have too much. I know he is huge in that field and he will walk you

Speaker5: [01:06:07] Step, step

Antonio: [01:06:09] By step in how to how to do that. And so I'll be happy to drop the link in in the in the comment. But, Tom, I see you have something that might be helpful for Mark.

Speaker2: [01:06:26] It's just that I am both the most experienced Google search. Skills I knew what terms to use to get to some good stuff. Basically, those three videos will get you a good start and then you'll know how to search for stuff better than me at that point. After watching the three videos, I have a friend to do this stuff a lot. If you run into problems, Mark, reach out to me personally with a LinkedIn direct message, and then we can go from there. I can hook you up with the people that really know the stuff well. Ok, sounds good. Thanks for all those links and suggestions.

Antonio: [01:07:03] Yeah, I think that's the best advice you can get when you're in the Data field. The number one skill you need to have is really is Google. And knowing how to look stuff up on that Golo or on YouTube. That's skill. Yeah. Like a lot of times they would ask me to work. They're like, hey, Antonio, can you do this project? Not only give me a second, I would it if it's on Google,

Speaker5: [01:07:25] I'm like, yes, I can do

Antonio: [01:07:26] It. No problem for sure. If if I can't find it on Google, then like I think I think I could do it. But it's going to take a lot of time. And you see the way everything is set up, it's going to take Doudou for Businessweek. So can we maybe we can work on something else. But that's that's the the the play

Speaker2: [01:07:48] Thing you need to do is I believe it's season four, episode four of Silicon Valley. You need to watch the hot dog, not a hot dog. That's probably the training [01:08:00] you'll get as a great show. Thank you for confirming, Monica.

Antonio: [01:08:06] I started that I need to finish up, but I can definitely relate a lot to it. Let me see on answered. I don't want you guys going to sleep with your questions unanswered. Did anybody help you or do you want to ask your question?

Speaker2: [01:08:31] Hmm. Hello? Yeah, I me like, is it just a matter of working? That's what I'm starting to like, watching some videos and like really getting into it like our what I call Gruebel system is working in our business. Like I'm not familiar with the global database. Like I'm like I know SQL and I have done work in using that.

Speaker5: [01:08:54] But Ndongo Debian

Speaker2: [01:08:55] Laws is like this platform and like I'm just trying to figuring all that stuff since. And this semester I have to do one Data. So what kind? I'm not sure like me since I got the first satellite is a like I don't have the system to play in my model locally. And you know, you don't have the resources to provide us. And I don't have that. I would debut on my leg, on my lap. So I was thinking like big my work on the cloud. So but I'm not very sure like I'm going to suggest like what cloud services like I can use or to like what other services I can use or train my models. Just one question I was going to recommend Clough's, and I'm going to need help here and it's to the left, but. My big error when I start buying my own cloud time is Data. I would do my best to minimize the time I was from that, so getting a girl Cimarron, I hope [01:10:00] I'm saying that, right? Yeah. Good. Good. Play with a toy sample until you get it right. You can do that in Google CoLab. Once you get that right and you're pretty confident you could scale the Data and everything's going to go smoothly, then go rinse cloud time. But I always pucker the most when I have to run Data thyme. We keep going over your hand, right? Yeah, let me keep that.

Speaker6: [01:10:30] Yeah, just I guess, I guess two quick questions. So what are the remnants of that act like? Did he tell you what it needed to accomplish that he tell you? Sorry not he did. They did they tell you what you need to do?

Speaker2: [01:10:45] To restore like specifically he's like you can do like research where you can like have your own nobility on it, like something that is like well before all you've been vote on the model. Again, they use the new machine learning to like use like just like the and the rest they use that you can do OK. Or do anything they use big data and then use like Elasticsearch to scale that,

Speaker5: [01:11:18] Then

Speaker2: [01:11:20] You some a bunch of technology like our vital and this is kind of. Ok. And.

Speaker6: [01:11:30] Know what any of these things are. And I find reason I'm asking that is because I want to understand kind of like what's your prior experience with either engineering or Data science and analytics?

Speaker2: [01:11:39] Yeah. So I'm worried like I did my bachelor's in computer science and often I have this research and done so. Looking at the research done. I vote on a machine

Speaker5: [01:11:51] Like prediction of this Philmore

Speaker2: [01:11:53] Using machine learning algorithm like it was a paperboys and I would be based device, which was you six different factors [01:12:00] into account. And we were using it to predict the soil moisture so that if like if rain is like about become

Speaker5: [01:12:07] Like three to four days,

Speaker2: [01:12:09] Then we don't need water, the proper gardening. It was the basic concept on it, like we wrote on it. And after that, like I'm just working on small projects, my prediction on some serious prizes. And I joined as JSON as JSU in in 2011. And initially, like when I first met, I was doing some Bovery stuff. And I and in semester I have like actually started working on machine learning like last working on a German model like I did in like from scratch brain like using the code for the Data and like to write on the beat about and building the algorithm from the scratch so that I get the inside codes. Look.

Speaker6: [01:12:52] Ok, so it sounds like is that for the project, you could correction's, you could either develop a new model, you could potentially production AIs a model, you could keep in a specific area, or you could do something that's a little more integrated. But basically, your professor gave you guidance. Ok, that's fantastic. So and then in terms of your like, would you say you're stronger in engineering or modeling, like you've had more experience with modeling? We are

Speaker2: [01:13:25] On red. I have like yours, Martin said that I have.

Speaker6: [01:13:38] You think our experience and I think in that regard so. I guess just for anyone else, it's kind of like going through something of war. So one thing I would say is for projects, because they're a great base, you kind of don't want to sign yourself up for too much. And for any scope, you assume your project is going to take an additional 20, 40 percent the scope, [01:14:00] because honestly, when you're learning a new skill set, that's about how long stuff takes is like more than you think it is. So I feel like in some ways I would just try to go for kind of the easiest minimum viable product. And one of those things I would recommend is looking at some example resources out there. So maybe with email is a really good resource. He walks through what a VP like model and production could look like, and it touches on the different elements. It touches on like personal self rendering skills, like being and logging and writing in Python, which on how do you like deploy? How do you monitor all these all these things for class projects that you ever need to go that much, that big? But it can be something as simple as like develop a model, try to get a hostess place and have like a store or it could be even something like a little bit deeper.

Speaker6: [01:14:56] Where? I don't know, you're trying to figure out the best way to AIs like a storage solution for Twitter or something, like how would you like stream it and then how would you store it and how would you analyze it? But I always feel like that's a lot. So I would say like look at me with email as like a really good overview of sort of like what? Big Data meaningless term project could look like within another series like Good. So there's a website called Confetti Dot III, which houses. Like machine learning data science, like tester exams or practices. So what you're trying to interview, they have all these like different questions that you can kind of like yourself on. They had this one like six part series, which was essentially like, how do you create a Crome

[01:15:42] In that identify as fake meat?

Speaker6: [01:15:46] So, yeah, one was a really good one. So if you do like fake new plug in like example with confetti, I look for them a blog post. That one was really nice because first off, like a lot of these Apple products I see they usually use streamlet [01:16:00] apps, which is nice. I think also that since this one is going a step further, where the guy relative to Orial and how, for example, if you're on a news article and you create this Chrome plugin, it'll show you it'll highlight the text and give you. Like a classification score, it's fake or whether it's fake, like any of those kind of like a dummy. It was a version which honestly is the best part. It was still like, I think I really, really well done example. So I would recommend for anyone who was starting like a big Data or machine learning information project to look at those two resources, for an example. And then also this scale down from there. And usually going with cloud provider is different because you can always get like student rates, for example, has like a freemium sort of offering. Yeah, and honestly, what the providers like their manager says of this, whether you go with Azure or of us, it really doesn't make a difference as long as you stick with that cloud provider like you don't want to be mixing it all the people just use GCP or use HWC or use Azure and understand that when you graduate and get a job anyway, they're going to make you use a different cloud provider and you're going have to learn it again. So it's just like my just real life

Antonio: [01:17:24] In delineating just as a tool and just go with it. I think a lot of them are. I mean, all of them kind of do similar functions. You pick up one, you'll be able to kind of go back and forth if your job allows it. But more Simran, I know you you drive for a little bit. Mikiko offered you some some great advice. After this call, we're going to be publishing it on the artists of data science and Harpreet Sahota publish it on LinkedIn. So I'd definitely recommend going back and listening to it. And you always like messengers or find somebody else? Linkedin. Sure, [01:18:00] Mikiko wouldn't be opposed to helping out if you have any additional question. But I don't want to sign Mikiko for anything. But I know she's she's nice. So let's see if, Eric, you have your hand raised.

Speaker2: [01:18:14] Yeah, I just wanted to throw in that. I don't know how you say there's some run, but I've got some Bhante. If you want to talk about this like this is

Speaker5: [01:18:22] This is like a part of like stuff

Speaker2: [01:18:24] That I've just had my experience in and I love to talk about. So, like, if you want to like if you want to, like do a screen share and we might make a diagram together, like at a time I'd be down to help you with some of the stuff for maybe scoping out what's what's reasonable to do. Anyway, so

Speaker5: [01:18:40] Larry, would love like

Speaker2: [01:18:42] You want to do like afterwards. Yeah. Yeah, like that's something special. Thank you.

Antonio: [01:18:49] All right, this is what we're here for, guys who we're connected and stay in touch and help me help each other out. Well, this has been a lot of fun. We're almost at the top of the hour. I want to thank everybody, the loves being here and gave me a hand when her is busy. Give them some good feedback if you think it's good job. Maybe he'll invite me back or maybe he'll band will never be allowed on here again. We'll see what happens next Friday. And a great man always says, you got one minute. Why not try to make big? So thank you, everyone. Thank you for dropping in all the fun and enjoy your extended the weekend those who have Labor Day off so. And area, I

Speaker2: [01:19:42] Think

Antonio: [01:19:43] You will get your free books in mid to end when they bobs like preordered. This is they cut in. I have your guys names. I will be definitely sending that out. So thank you.