HH50-17-09-2021_mixdown.mp3-from OneDrive Harpreet: [00:00:06] Let's go. What's up, what's up, everybody? Welcome, welcome to the artist Data Science. Happy hour. It is Friday, September 17th. I'm super excited to have all the guys here. Hopefully you got an Harpreet: [00:00:17] Opportunity to tune in Harpreet: [00:00:19] To the episode I released earlier today. This one was with Birdie Senga. The complete man had a great conversation with him. I really enjoyed chatting with him. Thank you so much for coming on to the show, man. Appreciate you joining me! Big shout out to Ken's Nearest Neighbor podcast. He had Vin VyStar on the podcast. I got to listen to a little bit of it, man. What a great episode. It's always good when those two get together, man, especially here in the office. I haven't seen this guys in quite some time. Hopefully AIs come back. It's been a while then and we're all at big shout out to everybody who's helped me with this course that I'm creating have had a lot of great people review it. Harpreet: [00:00:59] So the employable Harpreet: [00:01:01] Data scientists, that's the name of the company. I'm thinking the course is going to be titled the Data Science Mindset, but big shout out to the people that help review it. Tom Greg, Venn, Marc A. Harpreet: [00:01:15] Austin, Matt. You guys have been helping me Harpreet: [00:01:18] Tremendously, Elizabeth as well. Thank you guys so much for helping make that course. Something special. It's going to be awesome. Yeah, man. If you guys are joining us on LinkedIn, if you're joining us on YouTube or Twitch. Don't be shy. Come and join us in the Zoom room. There's a link to go ahead and get into the Zoom room right there on the comment box or wherever it is you're watching. Harpreet: [00:01:41] Happy hour number Harpreet: [00:01:42] 50 Man, 50 week straight. Been out here doing this Harpreet: [00:01:45] Thing in two Harpreet: [00:01:46] Weeks. It's going to be the one year anniversary of this happy hour thing, Harpreet: [00:01:50] Which I think is crazy. But doing this Harpreet: [00:01:53] Every Friday for an entire year, been here helping all out. Hopefully you've enjoyed it. We've had, [00:02:00] you know, started off real small. It got really big and busy for a while. That's small again, then big and busy. So hopefully we can keep this thing going. Super excited that that you guys have stuck with me all these weeks this entire year. So let's start off with an interesting topic. What I think is an interesting topic. We're having a little pre discussion Harpreet: [00:02:21] In the Zoom room before we got, you Harpreet: [00:02:24] Know, things kicking off. Tom was having a little bit of trouble with his, Harpreet: [00:02:28] With his audio setup, and Harpreet: [00:02:30] He's blaming it on just getting re acclimated to the Linux environment. So I'm wondering, man, all you guys out there, Data scientists Harpreet: [00:02:38] In the room in the chat. Harpreet: [00:02:40] What do you guys use? Are you guys Mac or Linux? Are you Windows and why? I need explanations. I need reasons, Tom go first. Speaker3: [00:02:49] Yeah, I need to clarify. I didn't realize my mic was low. And then when you told me it took me five seconds to do on Linux, what would have taken me maybe a lot longer on Windows? I used to have this old picture. Harpreet: [00:03:06] It was funny. Harpreet: [00:03:07] It was just Speaker3: [00:03:09] Really bad emoji type pictures, but one was someone bemoaning because they had to pay for an update. The same was true on Mac, Harpreet: [00:03:23] And then the Linux guy saw an Speaker3: [00:03:25] Update notice. Oh, more free stuff and it just updates for free. Even Linux Mint, which I love to use and it's just I've been using it for seven years now and it is just steadily gotten better every year. So its Mac is basically Linux. You pay for it, which I guess that's OK. Harpreet: [00:03:48] Yeah, yes, that's a good point. What about dressing room code stuff? Good to see Harpreet: [00:03:53] Here, Russell. Harpreet: [00:03:54] A.a. is not feeling too well, but coach, what about you, man? What is it that you use? Speaker3: [00:03:59] Yeah, the [00:04:00] first off, I'm happy to be here. I mean, last last week, thanks to all of you guys, you guys answered a fair few of my questions last week on Quantum. But I guess on this topic, I've recently become the main fence sitter on this kind of topic because I grew up with windows, right? I grew up like at home with Windows didn't really code until I was at university, and that's when I started picking up Linux systems and things like that, right? I find and I was pretty much anti Mac until I joined the company I'm at right now where we all use Mac, right? So it was kind Harpreet: [00:04:38] Of forced down my Speaker3: [00:04:39] Throat a little bit, but I found it surprisingly easy to use and I can see I can see different benefits to different systems, right? So with Linux, Harpreet: [00:04:48] You get that full scale kind of freedom. Speaker3: [00:04:52] So if I had to work on a local machine and I didn't have cloud access and I had to work on local Data with my own graphics card, Linux all the way because I have full control over how the graphics card interacts with everything else I've got. Harpreet: [00:05:05] You know, Speaker3: [00:05:06] I've worked in a defense industry job before where I'm doing, you know, object detection and stuff for for underwater robotics. So you can't have cloud access. There's no point there's underwater, Harpreet: [00:05:18] A robot can't Speaker3: [00:05:19] See. You don't have internet, right? Harpreet: [00:05:21] So in that situation, Speaker3: [00:05:22] You need to learn to have pretty tight Harpreet: [00:05:24] Control over your interactions Speaker3: [00:05:27] With your graphics card. So Linux by, you know, no doubt beats that all ends up. Harpreet: [00:05:32] But when you're Speaker3: [00:05:34] Talking about maintainability across a larger workforce, let's say you got 20 or 30 Data scientists or machine learning engineers, and they're trying to deploy a solution to, you know, and it's all cloud based solution. So US GCP based solutions. I've found it that I spend no Harpreet: [00:05:52] Time at all managing Speaker3: [00:05:54] My personal machine or my environment right, other than basically package installation, whereas [00:06:00] previously on a purely Linux Harpreet: [00:06:03] Like I had a Dell with like Linux, Speaker3: [00:06:05] Ubuntu Eighteen put on it, I spent a lot more time managing the, you know, the OS itself so that I could get to work. In this case, I'm not having to do that at all. I just kind of log in and GCP gives me all the graphics power. So there's that kind of trade off in terms of how much individual it support you need. Whereas if something goes wrong, my laptop now I can ship out a brand new Mac and then just log into that in about 15 minutes, I'll be up and running again. So there's that benefit to using like a more Harpreet: [00:06:33] Managed service, Speaker3: [00:06:34] Like a Mac or Windows machine. And I see that that's a little bit better with Harpreet: [00:06:37] Mac, and Speaker3: [00:06:38] The preference for that is because Harpreet: [00:06:40] The all the hardware Speaker3: [00:06:42] Is really Harpreet: [00:06:42] Packaged in right. Speaker3: [00:06:43] But personally, I actually use a Windows machine at home because I can I mean, I can change out the graphics card any time I like. No problems. Plus, I don't have to deal with the operating system level details like I would on a Linux machine. I can still use subsystem Linux if I Harpreet: [00:07:02] Want to use my Speaker3: [00:07:03] Full terminal. Harpreet: [00:07:04] So the latest subsystem Speaker3: [00:07:05] Linux seems to have most of the terminal commands that you can usually want. I'm just waiting Harpreet: [00:07:10] For in the premiere Speaker3: [00:07:12] Mode of sorry, not in the preview mode of Windows. They've got the graphics card, the Nvidia plug ins working with well. So I'm just waiting for that to come into one of the more stable releases. And then to be honest, I I reckon there'd be once we get to that stage. There's as much value in running Linux internal to Windows as there is running fresh Linux. That's kind of where I sit. Harpreet: [00:07:37] Thank you very much, Russell. Matt, what about you guys? What are you guys into? A. Here in the chat says that he is windows Harpreet: [00:07:46] Just because his specialist Harpreet: [00:07:48] Work computer is as he used Linux and Mac for other stuff, but nothing against them. Harpreet: [00:07:55] Yeah. E-mail AIs. Speaker4: [00:07:57] Windows, much like A. does because it's [00:08:00] my my employer's primary Harpreet: [00:08:03] Ecosystem, so Speaker4: [00:08:04] I use that a lot. I have a Mac at home, but I tend to use that just for a like a music server, basically, and a home entertainment system and, you know, iPhones, iPads and stuff that I use for that. I don't really do too much coding within the Mac system, so all of my coding is done in the Windows system. Harpreet: [00:08:25] And yeah, you know, I get Speaker4: [00:08:26] Frustrated with it. You probably see at the moment if my camera Harpreet: [00:08:30] Is playing up on the Speaker4: [00:08:32] The call as it is for me here. Harpreet: [00:08:33] Yeah, something Speaker4: [00:08:34] Between my my video card and the laptop and my GoPro webcam doesn't seem to like each other. So it kind of paints me out of the picture every once in a while, which is to everybody's advantage, I expect. But yeah, I'd like there to be far more Harpreet: [00:08:51] Capability for Speaker4: [00:08:52] Configuration within the within the Windows ecosystem, for the graphics Harpreet: [00:08:57] Cards Speaker4: [00:08:58] And the wider coding capabilities as well. We are not really using Linux, so I can't comment on Linux. Harpreet: [00:09:07] Antonio, what do you guys use? Mac, Windows, Harpreet: [00:09:09] Linux where you guys prefer? Speaker5: [00:09:11] Oh my god, I was literally just messaging my my Harpreet: [00:09:15] Wife because Speaker5: [00:09:16] So I started this week. I started a job at Google. Harpreet: [00:09:19] That's congrats. What are you doing over there? Speaker5: [00:09:21] Thank you. I'm going back to my roots. I'm doing fraud stuff. So fraud analytics and trust the safety department. Yeah. So this was the first week they gave us. They gave us an option of like a Chromebook, a Harpreet: [00:09:36] Macbook Pro Speaker5: [00:09:37] And like some unknown stuff. And everybody I asked, they're like, Do not, don't get a Chromebook, go with your MacBook. And because I'm always like, I was always like an HP user or Dell, and I got a MacBook Pro, and I literally just messaged my wife when I was trying to join this from my HP. I'm like, Oh my God, I don't want to. I don't want to use HP anymore. So [00:10:00] I was very surprised, but I paid Google. I think like from what I've. Like, 60 to 70 percent of like Data people and even other Vietnam like engineers there, they're very big on Max. So I think it's been a week only, but it's won me over the MacBook Pro. Harpreet: [00:10:21] Yeah, dude. Like I was a lifelong Windows user up until I was about thirty five years old. I started working at bold commerce and then switched to the Mac, and I have not looked Harpreet: [00:10:29] Back ever since. Harpreet: [00:10:31] I've been die hard Apple Mac ever since then. It was just very, very difficult to do and learn Data science on windows. I felt everything kept breaking. Everything kept breaking. When I was doing stuff in windows, I get all these weird error messages and whatnot. Harpreet: [00:10:48] Then once I Harpreet: [00:10:48] Got to Harpreet: [00:10:49] Ta ta ta bold Harpreet: [00:10:52] Here in Winnipeg, working with a couple of really smart people and they taught me how to use the Mac, it was a quick learning curve. Harpreet: [00:10:58] Super easy. Harpreet: [00:11:00] You know, I loved working out of the bash terminal and it just it was so easy to use. I haven't looked back since then when I went to price industries. They, for whatever reason, wouldn't give me a Mac if I had to use a windows. And the first thing I did was immediately get Wfll installed. And so I did all of my computation and data science work out of the subsystem. Harpreet: [00:11:21] The Linux subsystem, Harpreet: [00:11:22] Which was a Ubuntu two 18 at first and then up to 20, and it was just seamless and didn't touch any thing in Harpreet: [00:11:32] The Windows environment. Harpreet: [00:11:34] Monica, what about you? Harpreet: [00:11:35] What's what's your what's your go to device? Speaker5: [00:11:39] So I have I have a windows. I am trying to think if I have even ever done anything with Macs, I would be useless. I wouldn't even know how to turn it on. Harpreet: [00:11:53] Yeah, no, I have. I mean, Speaker5: [00:11:56] I've heard I've heard mixed reviews, though, because my, [00:12:00] my husband, he actually Harpreet: [00:12:02] He's Speaker5: [00:12:02] Working with Google and they gave him a MacBook. And so he was there's like just stuff that's a lot, lot different that he's still getting used to. But yeah, me personally, I have no idea about that world. I'm I'm all about my androids and my Samsungs. Harpreet: [00:12:20] And yeah, I guess that's one thing I like about Apple products is everything seems to Harpreet: [00:12:25] Connect together like Harpreet: [00:12:26] Everything my watch, my phone, my Harpreet: [00:12:29] Tvs, my laptops, all of Harpreet: [00:12:31] My laptops. Harpreet: [00:12:32] It's just it's amazing. Matt, good to Harpreet: [00:12:34] See you, Matt. How you doing? And by the way, if anybody has questions like we're just getting the conversation warmed up, I was just bounced around this question. Mac, PC, Linux just kind of get the ball rolling the conversation. But if you have questions, Harpreet: [00:12:43] Whether you are in Harpreet: [00:12:45] The Zoom room, whether you are in the LinkedIn on LinkedIn, I want your questions. I'm happy to answer your questions. Twitch YouTube. I would love to. Speaker3: [00:12:56] Real quick to question to the Mac lovers. I used Speaker5: [00:13:01] Macs for Speaker3: [00:13:03] Quite a while, many, many years ago and then. But having bounced around between Windows and Linux and Mac, Mac just seems to be a bit hyper controller. What do you say about Harpreet: [00:13:17] That Harpreet: [00:13:20] Hyper controlling Harpreet: [00:13:21] Like that? Harpreet: [00:13:22] I guess I haven't really Speaker3: [00:13:23] Like, for example, I'm doing something that I'd normally do just because I know what I'm doing, and the OS says we can't do that yet. You haven't done this and I'm like, Are you? It felt like being in grade school now. Do this and then do that, and then you can do that. I know what I'm doing. Quit telling me what to do. Harpreet: [00:13:48] Yeah, it's I haven't felt that now. I've been. I've been. I guess there's a lot of settings that you have to take care of in the settings before Harpreet: [00:13:59] Things start, [00:14:00] you know, Harpreet: [00:14:00] Getting getting smooth Harpreet: [00:14:01] And flowing. Harpreet: [00:14:02] Question coming in from LinkedIn Rodney. Have I changed my tech set up video is dropping in and out? Is that the case? It's my video being dropping it in and out, my friends. Or has it been pretty good? No, I agree. It's good. Right? Rodney might be good, man. Right on. Super excited to see everybody else joining you and Mark's in the building. Matt is here. Gio, what's going on? And then mark other mark mark Bartolo. Man, this is good. I'm happy to see you guys here. Matt Bratton says he's the only PC user at the company. Oh, that's interesting. Tell us more about that. Harpreet: [00:14:32] Do people do people look at you strange Harpreet: [00:14:34] As you walk around the office Robert is seeing pricewise which one gives you more bang for your buck, Mac or Harpreet: [00:14:40] Windows? I just like everything to be connected. Harpreet: [00:14:44] I'm just a huge Apple Harpreet: [00:14:46] Fan boy that, you know, price wise, Harpreet: [00:14:48] Bang Harpreet: [00:14:48] For your buck. Harpreet: [00:14:49] Probably go for like a Dell or an HP, but just integration into your life so Harpreet: [00:14:54] Seamlessly apples the way to go. Harpreet: [00:14:58] If anybody, I mean, unless anybody wants to chime in here, let's keep the conversation moving. If anybody has questions, please do. Let me know, madam. We're excited to take any of your questions. Again, big shout out to Tom and Mark Freeman Harpreet: [00:15:11] And Matt Blazar, who are in the room that helped me up the game Harpreet: [00:15:15] For my course. Matt and Mark, especially on Vin as well, gave me the most detailed feedback and he as well, and his feedback has been crucial. You guys, thank you. This course that I'm launching is going to be that much better because we got guys input, so I appreciate that. Jai, I go for it. Speaker5: [00:15:33] Oh, sorry. I have no question yet. Harpreet: [00:15:36] Ok. All right. No problem. What's going on, guys? Questions. Let me know. Let me know. Go for it. Coach Deb. Oh, you are my friend. Make sure you unmute yourself. Speaker3: [00:15:44] There I am. Muted of 2020, right? Harpreet: [00:15:48] You're muted. So I guess Speaker3: [00:15:50] With the Windows Mac question, I mean, the two two points is what's better price wise and do you find yourself locked in? I've only been on Mac for like two months now. The only [00:16:00] points where I felt locked in with a Mac is that in order to install a piece of software that I want, but I just Harpreet: [00:16:07] Use like, say, if I want to gear Speaker3: [00:16:08] Up for my Mac itself as opposed to using the web interface, right? I need it to have a Apple account to install that stuff. I don't need that with the Windows machine, and I sure don't need that with a Linux machine, right? So that bit of the ecosystem is pretty tied up, kind of to answer your question a bit. But other than that, in terms of actually doing the things that I want to do from a data science perspective or a machine learning Harpreet: [00:16:33] Perspective, I haven't Speaker3: [00:16:35] Found it restrictive in terms of, I guess, in terms of the question of the chat about price wise. Rob, I guess it really depends what you want, right, and what your business is set up to do at the end of the day. Right now, I'm working at a machine learning company where all we do is on the cloud. It's all cloud delivered solutions, right? So a Mac just becomes this out of the box managed solution with very low it overheads. On the other hand, previously I was working at a company where we were more engineering oriented, so everything was going on to robots. Harpreet: [00:17:06] And frankly, Speaker3: [00:17:07] We needed CAD software and we don't have the quality of CAD software on Mac and Linux that we do on Windows. Like you're just not going to see mechanical design software, electrical design software come to quite that level. Yet they're just there's no one putting that investment into Harpreet: [00:17:24] Doing that on a Mac or Linux, Speaker3: [00:17:26] And that's where it just becomes a no brainer that, yeah, Windows wins all those applications that comes down to what you need. Harpreet: [00:17:33] Thank you so much, Coach said. A lot of great comments coming to man. We kicked off a pretty strong conversation on this for like the first half an hour just talking about Mac, Windows and Linux. Matt Bryden says he has given Mac at the start, couldn't use power query without setting up parallels. Ended up directing ninety five percent of resources to the parallels. Machines finally called it quits. Harpreet: [00:17:51] He hated it. Harpreet: [00:17:52] Caitlyn says I love the smoothness and connectivity of the Mac and also I presume my voice. But Windows PC is easier to use when [00:18:00] it comes to software compatibility. Awesome. Thank you. We're going to go ahead and drop this topic, Mark. Go for it. Speaker5: [00:18:07] Yeah. So I have a pretty fun week out Harpreet: [00:18:11] Where we had a kind of mini Speaker5: [00:18:12] Hackathon, so well, low kind of pressure work, just trying to finish Harpreet: [00:18:19] Your project in a week. Essentially, I Speaker5: [00:18:21] Think we remember a while back I talked Harpreet: [00:18:23] About how I want to like, understand Speaker5: [00:18:24] Kind of like churn within our company and all these different components. So I use this hackathon to do it. And my question is just like, Harpreet: [00:18:32] How do you handle for Data Speaker5: [00:18:35] Projects that are just complete flops that just don't go the way you Harpreet: [00:18:38] Expected? You know, Speaker5: [00:18:41] Essentially, I think all this week I didn't build anything I was doing, basically troubleshooting so many things. And now my my deliverable is like, here's a landscape of all the things that are messed up. This is what we need to do to fix it, and we can unlock this value here, but we need to do all these other things first. So I have that component, but I'm curious how others handle like you push for idea or you have this project that everyone's super excited about. And then once you start, it's just roadblock after Harpreet: [00:19:14] Roadblock, and it's Speaker5: [00:19:15] Just a complete flop for for what it is. Harpreet: [00:19:18] I mean, I've had this happen, definitely. Harpreet: [00:19:20] But I just try to reframe it like instead of thinking of it as binary win or lose. I think of it as, OK, we have uncovered Harpreet: [00:19:27] Opportunities for us to Harpreet: [00:19:29] Do something to make it easier for us to unlock value somehow some way. Harpreet: [00:19:35] Right. So all these like Harpreet: [00:19:37] Roadblocks Harpreet: [00:19:37] And barriers. This is Harpreet: [00:19:39] This is great because you probably wouldn't have came across these roadblocks Harpreet: [00:19:42] And barriers had you not Harpreet: [00:19:43] Attempted this project. And who knows the implications further downstream of you taking care of these roadblocks and barriers like you could, it could help you unlock a lot of other stuff for the organization. Yeah, I had to. I've had to give presentations like, Yeah, I've spent weeks on this thing and I can't [00:20:00] do it. Harpreet: [00:20:01] But but. Harpreet: [00:20:04] To hear from Tom on this and then from Tom, let's go to Antonio. Speaker3: [00:20:08] So Harpreet, I know you'll love this thought from Dave and Andy Tracer Bullet. And if I may use some bad language, I'll go Cajun mode here. I love the term shitty tracer bullet meaning promise. You know, I get something that I can just release. I said, I promise what I'm about to show you a shitty. Harpreet: [00:20:31] But it will get better. It can't get any Speaker3: [00:20:34] Worse than this, but basically you start end to Harpreet: [00:20:36] End mark and then you thicken Speaker3: [00:20:38] It over time. That way, you know, you've at least kind of got the thread of what you've got to get going. Speaker5: [00:20:44] That's a good. I actually just read the chapter on tracer bullets in the in the book, The Pragmatic Programmer. So I really love that one, I think, but that's a good call. Harpreet: [00:20:54] I'll take Data. So like I have the Data essentially like Speaker5: [00:20:58] They're in different silos and the challenges bringing them in together to one source. And so I think one way I can kind of approach this just like pulling all the fake Data Harpreet: [00:21:07] Or not just pulling all the Speaker5: [00:21:08] Data put into excel sheet and say, like this will look like we brought it together. So I think that's a really great call out. Speaker3: [00:21:13] But another thing and you kind of prompted me, I put it in the chat start with really ridiculously small, simple fake toy data sets. That's another way to test your end to end, at least on your data flow. And it's even easy to create fake dirty data and fake noise. And. But then that way, you're not wading through the complications of our gargantuan dataset. That's hard to visualize, which you've learned to replicate the same kind of issues you will see, but at a toy scale? Harpreet: [00:21:50] No, that's a that's a great call. Speaker5: [00:21:52] So I think, you know, our hackathon ended this week. I spent as weekend kind of pulling in a little quick mock. [00:22:00] I guess like this wouldn't necessarily be a tracer bullet will probably be in the chapter of the prototype. Whereas like, this one looks like you Harpreet: [00:22:06] Scrap it afterwards, but it Speaker5: [00:22:08] Gets a start. Harpreet: [00:22:09] You know, Greg is a fan of a scrappy, scrappy solution. Let's go to Antonio, then Greg. If you got any insight or Monica after that, you guys got any insights. We can go Antonio, Monica, Greg and then by the way, if anybody has questions. Harpreet: [00:22:20] Go ahead and let me know Harpreet: [00:22:22] Right there in the chat, wherever it is you are. I will add your question to the Q. After Mark's question, I've got a question from coming in from LinkedIn from Mojeed. We'll see if we can answer that question or not. But if anybody else has questions, let me know Antonio. Go for it. Harpreet: [00:22:37] So first, I'll say what Tom said. I think because Speaker5: [00:22:41] Mexico is not here, so it's going to be what Tom said. But I think, yeah, I think it's very important to start that proof of concept, develop as small as you can and see see what they should do. I don't think there's anything wrong with saying Harpreet: [00:22:56] Like, this project looks like Speaker5: [00:22:58] It's going to take a lot more time than value is going to bring me, you know, I think it's very underestimated, but I used to kind of like push for those things like, Oh, I said, I'm going to do machine learning or whatever, and I have to go through it, you know? And sometimes the better thing is saying, this is going to be a lot of wasted time or resources, and I don't I don't see it going anywhere. You know, so sometimes having to, you know, like take a step back to move forward and really evaluate it if you're asking the right question. I know a lot of times when you reframe the question that you're trying to solve for, maybe something else will open up. Right. So then trying to brute force it and trying to like different techniques. So I don't know exactly right what your project is, but Harpreet: [00:23:43] Think Speaker5: [00:23:43] About how, why it's not going. You know, if you can ask the question slightly different or maybe group, if you have a multi classification problem, maybe can I narrow this down into a binary classification, right? Just an example. Or if I can do any supervised can I start with unsupervised [00:24:00] learning? So just trying to re reframe it and look at it from a different angle that might help you? Again, depends on what your situation is. Yeah. Just to get I wish, I wish I was dealing with Ml problem, but it's clearly Data access. It's we have four silos. I'm trying to understand that we have a customer health Data. So our product Data kind of like course and sales data bring that all together. But they live in like SanDisk and Salesforce and database and BigQuery. And so like bringing that all together into one source, it's interesting. And I think the key linchpin I basically dealt with like I have all these tables, but like none of them have organization ID. And so I'm trying to figure out how I get the organization I need to Harpreet: [00:24:44] Merge across things Speaker5: [00:24:46] Has been the linchpin for me for this week. Harpreet: [00:24:50] But going back to your point, Speaker5: [00:24:51] Like simplifying, it's like, All right, maybe I don't bring in these four tables the low hanging fruit of these two tables, and that's like a step further than before, right? And show them, show them what you can do. Be like, Hey, you know, and then maybe you bring some of a manual. That's what I used to do, because we used to work across different organizations. You bring it in manual and be like AIs. This is what's possible if we have this Data, right? But you can see you guys are not labeling it correctly. If you want what I've built for you, here is a prototype. Start labeling the Data for me. This is how you're going to do it and what I would do with non-technical teams. I would say I will try to make it as simple as they can if I have to be like, Just do this in Google Sheets. You guys like your excel. You like Google Sheets. So I will do a dropdown every time you do this row. This is where you're going to label it as. And maybe right now it's not possible, but maybe in a month or two or three months. Because if you guys do this correctly, this is what we can do. So kind of motivate them, inspire them, and ultimately it's on them, right? Because you can build the best prototype. But if people are not going to take the time to look at it a long term and give you good data that I think is just going to be more, more headaches for you long term, because then you're going to have to keep fixing it. [00:26:00] Harpreet: [00:26:00] You also teach them how to fish Speaker5: [00:26:02] And then you help them out. Harpreet: [00:26:04] Definitely. Speaker3: [00:26:05] Just a quick interjection, mark. Harpreet: [00:26:08] I've always seen the Speaker3: [00:26:10] Data frame tools like pandas and the pandas look alikes as being the great integrators. Are they not giving you help here? Speaker5: [00:26:20] No, I'm fortunate because it's less like the rest of the Data itself. It's more so the systems. And so I'll give you a great example. Like the Zendesk, we we pull our data into a database, use the Zendesk API. But for the use case, it was for like like creating tickets, but they never brought in the organization field. So I'm going to have to go into the code base and update the API Harpreet: [00:26:43] Call to bring Speaker5: [00:26:44] In that data. So it's almost like the Data engineering side. I'm like, All right, I need you create a call for an API. Put that in our database and then move that to BigQuery. And then for like Salesforce, we have five trained for that, which is really great. But like there's like various permissions for the account that are like missing. Harpreet: [00:27:05] It's not to like, go Speaker5: [00:27:05] Scramble and get the permissions right to get it in. So it's like more bureaucratic kind of things and like operational things. If I had the data, it was just no easy merge saying pandas or even just BigQuery kind of kind of thing. It's more so access. Speaker3: [00:27:24] Well, and I'm wondering to cats. Yeah, yeah, I get it. When I was learning powered the AI a while back, it seemed like it was a fascinating integrator of data streams, too. Is that a possibility for you? Speaker5: [00:27:39] I think BigQuery is probably going to be Harpreet: [00:27:40] The more more because I'm Speaker5: [00:27:42] Trying to go towards like the Lake Harpreet: [00:27:43] House format. Speaker5: [00:27:44] I'm trying to push my company towards that Harpreet: [00:27:46] Where you have the Speaker5: [00:27:48] Elt kind of format extract load and your transformations within BigQuery with SQL. Harpreet: [00:27:54] But yeah, Speaker5: [00:27:55] Currently everything kind of goes into BigQuery, and we kind of combine all the data into like curated [00:28:00] views like a Data mark. From there, what do you call? Harpreet: [00:28:04] I'm just going to make a quick joke like what do you call the person that operates it? Was there a name for that? Because because you're going to need a Data canoe to get from a Harpreet: [00:28:12] Lake house to the Data shore Harpreet: [00:28:14] To deliver Data value? So build them Data canoes. Sorry, go for it, Antonio. Monica, I know you're you were unbeaten, so I was wondering if you had any insights here as well. Definitely go for it. But yeah, go for it, Antonio. Speaker5: [00:28:26] Also, find further Harpreet: [00:28:28] Context that might Speaker5: [00:28:29] Be helpful. I'm in a startup Harpreet: [00:28:31] So like a lot of fracture Speaker5: [00:28:33] Things we all figure out and be scrappy as we go and we're slowly combining things together. So that's that's another piece of information I just wanted to add to like people who are listening on LinkedIn or want to be like indies in the Data field, like what Mark is describing, I think it's honestly like Harpreet: [00:28:51] Like Speaker5: [00:28:51] 60 percent of the job is like he's saying, if you have the Data, oh, then I will do it. But the problem is, how do you get the data all in one place, right? It's not like you go on Kaggle and you download it. So a lot of the data science data analyst, whatever you want to call it, job is the problem is getting the people to agree on the same problem and solving it and bringing all the data together. And so if you like like doing all the cool stuff, you definitely have to be OK with all this like dirty work. And I always say that like 80 percent of the problems I have at work is not like technology or Data related. It's always comes down to people once you get people to agree on something. Usually the Data part we can figure out, but it's it's getting like five or six different groups and 30 different people across the organization to say, Oh, yeah, let's this is what we're going to do. Exactly. So just wanted to add that. Harpreet: [00:29:40] Monica, did you want to chime in with Monica Costa or are you still here? Yes, you are. And then we'll go to Greg after that. Speaker5: [00:29:46] I just love that if I have the Data, I've been there so many times, but very, very common to have all of these different departments that you have to get them together and figure things out. One thing [00:30:00] that I've learned, though, it's really easy to get sucked into, like all of the problems and all of the what ifs, and I can't do this and I can't do this. Try to always come with like a solution of what you can do. Harpreet: [00:30:14] Or if we Speaker5: [00:30:15] Did this, then we'll be able to accomplish X Y Harpreet: [00:30:18] Z, Speaker5: [00:30:20] But everybody else, you know, the low hanging fruit, the BS, all Harpreet: [00:30:24] Of that really, really Speaker5: [00:30:25] Great advice. Harpreet: [00:30:26] But just focus on presenting presenting Speaker5: [00:30:29] A solution rather than being the Debbie Downer. I guess that I definitely present as a negative, but like you said, it's all part of the job. I think the only thing is like, typically, I just don't have a week to do this. It'd be like a long term project. There was a hackathon week. I was like, Yeah, I could totally try to build a model for fun Harpreet: [00:30:46] In a week. Speaker5: [00:30:47] And like naive me, I knew I knew when I said that like that wouldn't be the case, but you know, I thought, Oh yeah, the data's already there. Harpreet: [00:30:56] No, Monica's Harpreet: [00:30:58] Point. Like, there's, you know, the classic Harpreet: [00:30:59] Military line, either lead, follow or get out of the way. Don't be the person who doesn't want to lead, doesn't want to follow, doesn't want to get out of the way. They want to tell you why the thing's not going to work, but that's the person that you don't want to be any follow points on that, Monica. Otherwise, we can go to a coast to coast. Speaker5: [00:31:16] Greg, no, I love that explanation, though. Speaker3: [00:31:19] Yeah, look, I totally I totally agree with that. Like, get out of the way if you're not presenting part of the solution, right? But at the same time, there's always value in being very honest with like either clients or internal to your company when, hey, if it's if it's really, really, really stuck up and if you're if you're up the wrong creek, Harpreet: [00:31:38] You've got to be Speaker3: [00:31:39] Really upfront and honest with that. Otherwise you end up spending months and months, you know, just working on this stuff and just patching stuff up, and it's not going to be a scalable solution, right? But this kind of I mean, one of the things that Antonio touched on is is more selling that story of, Okay, if I can, if I can take this segment of it and prove this business value [00:32:00] right? You'd need you can convince a company to then go and invest more into bringing Harpreet: [00:32:06] That their Speaker3: [00:32:06] Data to a point where they can monetize it better. Harpreet: [00:32:11] And I think where Speaker3: [00:32:12] I'm seeing that really start to scratch the surface here is more on the manufacturing side of things Harpreet: [00:32:18] Now for the Speaker3: [00:32:19] Last like 20 years, right? And in the last 10 years, what we've noticed is there was this entire change in design practices where design for manufacture became like the key words. So you'd manufacture a physical product so that it can be so that it can be assembled, disassembled, repaired really easily. What we're slowly moving to is very much and we're just starting on. It is designed for data science, right? Harpreet: [00:32:45] So what if I Speaker3: [00:32:46] Had to look at my real world physical Harpreet: [00:32:47] Product and I had to be able to detect Speaker3: [00:32:50] Problems with it? I had to deal with the supply chain of it. I had to deal with assembly lines. Now a lot of that stuff is still very manual right now, and you need the Data in a cohesive manner. You need that acquisition with the right modalities, whether it's a visual, whether it's, you know, numerical data coming through, you need to think about how you store and use that data through a manufacturing process. So we're starting to see more and more companies become aware while they're trying to integrate data science into their manufacturing processes that their current designs like. For example, if you have a physical product that is indistinguishable from Harpreet: [00:33:26] Something else Speaker3: [00:33:27] That sits on it, and that's a super critical component, and there's no way that you can visually detect that. How else could you do it? So you need to start designing in a manner that it becomes Harpreet: [00:33:36] Easier for data Speaker3: [00:33:38] Science to slip into your stream and really optimize your manufacturing right for this design for manufacturing is kind of designed for data science. This is the new thing I think where we're moving towards. Speaker5: [00:33:48] I think you're so spot on because I think you basically describe my job for the past seven months is like my my OK. I always like improved data access throughout the whole entire organization, Harpreet: [00:33:58] Which like I was able Speaker5: [00:33:59] To [00:34:00] deliver on which is super hyped about and is now I'm just doing extra stuff around that. But like keeping it was like a lot of Harpreet: [00:34:05] Training, different Speaker5: [00:34:06] Parts of the company, especially engineering of like, how do we get logs? How do we structure logs? We have a no SQL database, so it's very nested and normalized. And so, you know, there's a lot of trickiness that we have to go through to massage the data into a useful way and serve it to the right people. Harpreet: [00:34:24] And so, yeah, it's Speaker5: [00:34:26] Totally just like I told Tom earlier, wrangling cats, that's what it feels like. I'm herding cats all day, Harpreet: [00:34:33] But slowly Speaker5: [00:34:34] I'm training folks around. I'm like, This is why this structure in this format is really. Ornate. So I think that's a really cool analogy with the design thinking, not design thinking, but like manufacturing design similar for Data. Harpreet: [00:34:47] Go for it, Greg. Greg, we'll go to. Speaker5: [00:34:51] I just wanted to add a little bit of Harpreet: [00:34:54] Project management Speaker5: [00:34:56] Aspect to it because, Harpreet: [00:34:57] You know, everybody Speaker5: [00:34:58] Here Harpreet: [00:34:59] Already gave some Speaker5: [00:35:00] Some great pointers at the end of the day. To me, it's about communication, right, driving visibility to your stakeholders. I think most at the time when we see roadblocks because I'm trying to remember what she said earlier is that you come into you, walk into roadblocks after roadblocks, after roadblocks Harpreet: [00:35:19] And Speaker5: [00:35:21] The biggest Harpreet: [00:35:22] Disappointment to a Speaker5: [00:35:23] Project manager is not being able to, I guess, plan for time when it comes to roadblocks and not creating that visibility to stakeholders because stakeholders what they want, they want to make sure they hear what Harpreet: [00:35:41] You know, they Speaker5: [00:35:43] Know they are in the know of how things are evolving. So it comes down to one thing communication and communication, and the project starts at the time where you've done all the scoping that you can do and you lay out the plan. You beef [00:36:00] it up with some sort of, I guess, room for error or room for trials, room for discovery. And also you lay down the plan Harpreet: [00:36:09] For not playing, but Speaker5: [00:36:11] You lay out the risk that is included inside of that project to because they want to know, OK, if I engage in this activity, those are the potential things that I may discover. This is the risk that I may incur. And this is the time that I'm estimating it will take. And as I go through it, I will let you know when things happen right and why and what I will do to correct them. And then I agree, hopefully with everybody Harpreet: [00:36:39] Here that with the mention of Speaker5: [00:36:41] Places, so proof of concepts are definitely the best approach where you start small and then you know, you deliver Harpreet: [00:36:51] In a phased approach. Speaker5: [00:36:54] Never go go big on those things, especially if these innovations they are new, which means you are going to discover new things that you didn't plan for. And that's the nature of a project, right? So any stakeholder who doesn't understand that you will come across roadblocks isn't the best stakeholder to work with, because by nature, especially innovative projects by nature gets blockage. It's the continuing of visibility, communication and collaboration that gets you to the next steps. So that's pretty much what I wanted to add from a project management perspective. That's awesome feedback. And I'm also super happy that like this is just a hackathon, so it's like no major deliverable tied to it or major stakeholder. Harpreet: [00:37:45] So if this is like a Speaker5: [00:37:46] Project project, I'll I'll be freaking out right now. Harpreet: [00:37:50] But you got I mean, if it was a project project and you actually failed, it wouldn't be the end of the world, right? I'm sure your coworkers would be understanding they'd be like, Hey, man, at least you figure this out this out of this out. But you know, don't [00:38:00] don't catastrophize failures again. Flip it, flip it from negative to positive. Speaker5: [00:38:06] Tom, hundred percent good call out. Speaker3: [00:38:08] Yeah, real quick, mark. So Russell and Harpreet: [00:38:12] I were kind of bouncing Speaker3: [00:38:13] Around this in the chat, but it seems like you need to find some common format that all Data sources could go to so that you could then draw from that format. Is that is that right? Speaker5: [00:38:28] It's been a six month conversation I've had and slowly chipping away to make that happen. I completely agree. Harpreet: [00:38:36] So just Speaker5: [00:38:38] This thing Speaker3: [00:38:40] In my experience, if you think about we can call it hash tables, JSON objects, no sequel, they're all kind of the same thing dictionaries. In my experience, you if you plan those out carefully Harpreet: [00:38:56] And you make Speaker3: [00:38:57] Them a little more detailed and ugly than they need to be, you can kind of prepare for normalization with those. And it seems like everything is willing to go to something like Jason. So maybe that's your bridge. Like in a no sequel sense, but where you're actually preparing unique identifiers, such you're preparing keys so that when you get there, I mean, that might be in a sequel so that as you're building that neutral, everything can go to it. Typekit Data structure. Then in the final step, that could be that can go anywhere else from there. Harpreet: [00:39:39] I'm just thinking out another. Speaker5: [00:39:41] Another quirk, another another quirks, is because I think you might find it interesting that we use proto buffers, which is like a really Harpreet: [00:39:48] Strange Speaker5: [00:39:50] Use. And so, like the forces you to use keys for things. But because they have no single structure, we have [00:40:00] these keys, but then also have keys like nested within values as well. Harpreet: [00:40:06] So it's a very interesting dynamic. Speaker5: [00:40:10] You know, again, start up, I think there are like Harpreet: [00:40:12] Decisions made Speaker5: [00:40:14] For a much larger scale company, something Data structure. And so now I just as the and they don't have data scientists, background or data analysts. So I think now like coming in from this Data perspective, it works really well for web apps like it's super strong for web apps, but for analytics, Harpreet: [00:40:33] I'm like tackling the technical debt to Speaker5: [00:40:35] To to make it make it work and we're slowly getting there. Well, I completely agree. Getting a unified kind of data structure that's that's kind of so key. And that's that's that's why I really love Databricks, Databricks, their lakehouse format. I've been really falling in love with that because I think it really solves a lot of the kind of issues that we face. Harpreet: [00:41:00] What's a protocol Harpreet: [00:41:01] Buffer? Break that down real quick or throw buffer, and then we'll go to Matt Matt Blass after that. Speaker5: [00:41:06] Yeah, let me let me try my best way to to describe if someone else knows how Harpreet: [00:41:11] To describe it. Speaker5: [00:41:12] Totally go for it, because that's I don't touch as much. But essentially protocol offers for my understanding is like this data structure that was made out of Google. And it's supposed to be a a way that's scalable and a really fast kind of data structure for your data. And the way it works is that you kind of set forth like your keys and the structures of the day you call it out where you basically have code mappings of it before you fill it in with data and you create like these objects Harpreet: [00:41:44] With that and then through Speaker5: [00:41:45] Those objects, you fill it in with data and so you'll you'll create the mappings and then you run a Harpreet: [00:41:50] Compiler to like, create this Speaker5: [00:41:52] Auto generated report. And then from there, you fill it in with data within your database. It's beyond me. Harpreet: [00:41:58] I'm still learning Harpreet: [00:41:59] About [00:42:00] plans to go for it. Speaker5: [00:42:01] Yeah. Mark, could you tell me Harpreet: [00:42:02] More about what you were Speaker5: [00:42:03] Doing with Databricks? Speaker3: [00:42:04] Was it proto buffer? Speaker5: [00:42:06] No. So I was. We don't use Databricks Bazemore. So Databricks is the first one to really come out with the whole lakehouse paradigm. And they have this blog where they detail what were the 10 facets of like using a lakehouse and so Data breaks. You can't build a lake house with it, but you can use other different, different kind of vendors as well. And so the idea essentially is that like what data lakes it became like Data Swamp, because all your data was there, it was hard to understand what's happening. And then with data warehouses, they were just kind of like too structured and hard to use for like the different types of data that's kind of proliferating now. And so with the push now with cloud computing, essentially costs for storage dropped dramatically and the speed also increased substantially. And so this lake house format, where kind of brings the best worlds of a lake data lake and a data warehouse together in one format. Oh, OK, yeah, because I'm trying to learn more about like data bricks, and it's been a big curve. Yeah, I haven't used data bricks before as more so like they're the ones that came up with the framework through their blog. Ok, I'll check it out. Thank you. Harpreet: [00:43:17] Adam, big shout out we got Matt Brattan in the house, Matt. Good to see you. Matt is the purveyor of amazing analytics shirts. If you see me rocking my sequel, that's like from a shirt. Harpreet: [00:43:27] That's that's Matt doing right there. Harpreet: [00:43:31] He's also got this push button. Get analytics. Harpreet: [00:43:33] I love that Matt. Drop a link to to your Harpreet: [00:43:36] Shirts right here in Harpreet: [00:43:36] The in the chat. Harpreet: [00:43:38] Big shout Harpreet: [00:43:38] Out. Also, Matt, you crushed Harpreet: [00:43:40] And surpassed five K on LinkedIn. Harpreet: [00:43:45] That's huge. Also, you know, big shout Harpreet: [00:43:47] Out to Eric Sims past that 10k mark. All right, let's go to a question from LinkedIn. All right, so we've got a question from LinkedIn coming from Madjid. He's asking What should [00:44:00] I do to make my master application solid for master data science specifically if my undergrad was in my ass, which is a business degree? Any tips? So we're going to have to make a lot of assumptions here Harpreet: [00:44:13] For the Harpreet: [00:44:14] Purposes of this conversation. Let's just assume that you are in North America applying to North American schools. For what it's worth, let me just give you a little bit of history of Harpreet Sahota. So I graduated with my undergrad from Cal State Fullerton at like I studied economics and math, and Harpreet: [00:44:31] My GPA was like barely enough Harpreet: [00:44:33] To get a degree I need to point to. Literally, it was like a two point two GPA somehow because of Harpreet: [00:44:39] California's Harpreet: [00:44:41] Really high bar for educators. I became a teacher of mathematics. Harpreet: [00:44:46] Somehow I took that there's this series of exams. Harpreet: [00:44:49] I had to take math. But I did well on those, and then I wanted to get back into grad school, it's like, you know what? Teaching is cool, but it's not really going to get me where I want to go. I should become an actuary because then I'll make a lot of money. So I started applying for schools. No graduate program would accept me because of my GPA. So I had to go to UC Davis and UC Davis wouldn't even let me in as like, you know, Bachelor student. At first, I had to take a course after course as a what was called graduate student at large. And so I took pretty much every undergraduate statistics course I could at UC Davis and maintain like a three point seven GPA during that time. Then I took the green. I scored like in the 90th percentile for math and then just wrote a killer essay, and I got into a bunch of different grad programs. Harpreet: [00:45:40] So that being said, if my dumb ass could get into grad school, so can you. Speaker5: [00:45:45] So if you Harpreet: [00:45:46] Are going for an app, you know, most Harpreet: [00:45:50] Programs will have a pre Harpreet: [00:45:52] Master's program, right? And that pre master's program is designed for you to Harpreet: [00:45:56] Catch up on some Harpreet: [00:45:57] Foundational courses that [00:46:00] are going to be necessary for your success and further graduate courses. You can either do that on your own at a local community college or go through the university's Harpreet: [00:46:10] Normal grad program. Harpreet: [00:46:11] But again, we're making a lot of assumptions here Harpreet: [00:46:14] That you are in the U.S. or in North Harpreet: [00:46:16] America applying North American schools. Harpreet: [00:46:18] So if that's the case, hopefully your GPA Harpreet: [00:46:22] Wasn't as low as mine in undergrad and you know you got to if your program requires a degree, I just make sure you crash the degree. Do a couple of side projects actually get in Harpreet: [00:46:33] Touch with Harpreet: [00:46:35] Some instructors? So I made it a point to reach out to. So I applied to a bunch of different schools and I made sure I reached out to the head of every single department, said the quick email. And just ask, like, Hey, what is your ideal graduate student look like? Harpreet: [00:46:47] Like, what do you want? Like, what Harpreet: [00:46:49] Kind of students do you want in your program? Right? So try that route as well. Look up the schools that you really want to go to find the department heads, send them emails, you know, find the subhead or wherever they're called, and send them emails and just get a Harpreet: [00:47:04] Get an idea of, you know, hey, Harpreet: [00:47:05] What? In your opinion, what does it take for a graduate student to succeed in this program at this university? Thomas said, You had your hand up, Harpreet: [00:47:13] So if you want to share some Harpreet: [00:47:14] Advice, definitely go for it. Speaker3: [00:47:15] I want to be here. The question was it I thought he was almost asking, but should he study in his master's program? Harpreet: [00:47:24] Now he wants to research. Harpreet: [00:47:27] No, no. What should I do to make my master application solid for a master's in data science because my undergraduate was in MIS, which was a business degree? Speaker3: [00:47:38] Yeah, I think you were hitting on all cylinders there. Harp. Harpreet: [00:47:41] Mark, go for it. Speaker5: [00:47:42] Yeah. So I, Harpreet: [00:47:45] You know, at the Speaker5: [00:47:46] End of the day, the way I view application, especially to graduate school, it's a negotiation you're trying to negotiate with your counterpart Harpreet: [00:47:53] That like, I have this Speaker5: [00:47:54] Value that you want to bring into the school district, especially for competitive programs. So [00:48:00] for me, like I'm still shocked that I got into Stanford for for my masters because similar to you, like a master's in science program, I signed a two point three. My overall GPA was barely a 3.0. My background was sociology. And so essentially what I did was, you know, I worked really hard ahead of time to to get Harpreet: [00:48:24] The the personal Speaker5: [00:48:26] Statement down as well. And also, I did the Jerry like where were the average was I was like one percent above average, so not stellar at all for these things. But through that, the one key thing which I think got me in was like, Is there anything else you want to tell us Harpreet: [00:48:43] That that thing can be gold Speaker5: [00:48:45] For your kind of application? Harpreet: [00:48:49] Because what I Speaker5: [00:48:50] Essentially did was I called out my flaws and own the narrative around it, so they couldn't. And I said, like, these are my flaws, and this is why they're huge shrinks. So I basically said, Hey, my grades sucked. Here were the situations around that and the lessons I learned. And because I went through these lessons, I essentially talk about imposter syndrome and how I like avoided studying because I did leadership things as a way to fill this void of, like completely failing in school over and over again, right? And how that created the cycle of like, Oh, I didn't do well in school. I took on more leadership opportunities, which prevents me from doing good in school and created the cycle. And so essentially, I called that out and I said, Hey, research was the first time I felt like a good student and it broke that cycle. And this program is a research program which aligns with my passion and because I went through these lessons. Harpreet: [00:49:47] I won't do it again and imagine me. Speaker5: [00:49:50] All my leadership experiences putting all concerted effort into this program, I would be a huge asset that you'll become wild for you, miss out on, right? Harpreet: [00:50:00] So [00:50:00] I took the main weakness on my Speaker5: [00:50:01] Application and just highlight it as a shrink. I'm happy to share that that personal statement with anyone. I share it to a whole bunch of people all the time. I'm happy to make them more public. If you want an example of how to to navigate that kind of tough conversation, but that was the key thing that really Harpreet: [00:50:20] Kind of put my Speaker5: [00:50:21] Application over the edge. Harpreet: [00:50:23] Yeah, absolutely love that. That's what it's called personal statement. That's the graduate school essay. Harpreet: [00:50:27] But yes, what Harpreet: [00:50:29] Is so awesome? What you said reminded me of like that last rap battle from Eight Mile, where Eminem is just like listening to all his flaws and like, All right now, tell me something about myself. I don't know. Harpreet: [00:50:39] I love that absolutely, man. Speaker5: [00:50:40] And no need to say something real quick and told You need to call you. What Mark said also applies in the real world. You will not run away from making mistakes, but Harpreet: [00:50:51] There is power and Speaker5: [00:50:52] Owning up to your mistake, explaining what you've learned and what you will do going forward. Keep that for the rest of your career. You'll do well. Harpreet: [00:51:01] Very good. Thank you so much. I appreciate that. Yeah, great. Great advice, Antonio. Go for it. Speaker5: [00:51:06] So when I was applying for graduate school, I noticed the one thing that the schools really want to see and you touched upon that is taking a couple of classes Harpreet: [00:51:16] From their school. Speaker5: [00:51:18] So like, I had solid grades, I had a really good GPA and everything, and I applied to Georgia Tech Harpreet: [00:51:27] And they just denied me and I'm like, Speaker5: [00:51:30] Ok, what do I have to do? And they're like, Well, you know, if you could take Harpreet: [00:51:33] This, there's Speaker5: [00:51:34] This micro masters at X and you think these couple of classes, I took those couple classes. It was like so much easier than what I had done in my undergraduate, but I didn't care Harpreet: [00:51:45] Because that's what they Speaker5: [00:51:46] Cared about. And as soon as you as soon as you do their classes and you do good like you get in like a or B, they just kind of like, I guess to them, it proves like, OK, this person is up to up to par with this [00:52:00] degree. For the record, I didn't end up going to Georgia Tech because once I took the classes, I'm like, Oh, this is not for me. So that could also be a thing where you take the classes and you're like, Oh, maybe this wasn't the the thing I should have been like aiming for. But I know, like Michigan has a lot of data science programs. So if you're looking like Michigan school or I, maybe you can do that on Coursera or Georgia Tech has most schools now have some kind of online programs that you can you can start with. So definitely, definitely go for that. Harpreet: [00:52:29] Thank you very much, Antonio. Yeah. For the record, I don't think you need a graduate degree to go to get a job and data science like anybody can learn it from all these massively open online courses. Do projects that you can get a job without a graduate degree. That being said, like I looked at the guy's profile. You're and you said it in Harpreet: [00:52:47] Linkedin Harpreet: [00:52:48] Here that you're trying to get Harpreet: [00:52:49] To Canada and you're right now Harpreet: [00:52:50] In Egypt. So yeah, definitely. If you're trying to, you know, get a better life for yourself at a different country, then sometimes going to school is the route to do that. Harpreet: [00:52:59] That being said, Harpreet: [00:53:00] Man, let's go do a Matt Damon Matt, whether you got another question or if you Speaker5: [00:53:03] Know why. Yeah, I just want to pick back up on that. Continue to Harpreet: [00:53:07] Thread what Speaker5: [00:53:09] Everybody here has advanced degrees in Data science or something of the data science ilk. What did you learn in that program that you would not have gotten through equivalent tangential data science experience Harpreet: [00:53:22] That you can enter a little quote Speaker5: [00:53:24] Unquote Data science world? What what's the value out of a grad program? Harpreet: [00:53:28] In essence, like I've learned Harpreet: [00:53:31] Far more outside of grad school than I did while I was in grad school, but if anything, Harpreet: [00:53:37] It forced me to Harpreet: [00:53:39] Learn the Harpreet: [00:53:39] Fundamentals at a very, very Harpreet: [00:53:41] Deep level, and it taught me how to do research. Those are two things. I mean, you can still do that on your own. You just you have to be sufficiently motivated to do that. But it helps to have a rigorous structure Harpreet: [00:53:54] And program like Harpreet: [00:53:55] In place. I'm to turn it to the rest of the audience that let's go to a coast [00:54:00] of. And then after coast up, let's go to Marc Freeman Harpreet: [00:54:03] And just to shadow Speaker3: [00:54:05] Ben Taylors in the building. Yes, I mean, I think fundamentally, I finished my master's degree just over two years ago now, so I'm still pretty fresh out of the grad program system. But what I was looking at and I'm from Australia, so I'm not situated in Canada, I'm not situated in the USA where I have access to, you Harpreet: [00:54:23] Know, all of those things. Speaker3: [00:54:25] So I need it to do Gree before I could apply to any American universities. I ended up going to a British university instead because by the time I finish a degree, go to an American university for a couple of years, finish that off. I'll be two three years down the track in my career, right? What I basically got down to was I spent a bunch of time learning whatever I could on my own and got to the point where I needed that bit of guidance, right? And you've got to be really self-aware of. Okay, this is what I can teach myself. And then this is the thing that I'm struggling to learn for myself, and that's where, you Harpreet: [00:54:56] Know, for lack of a better Speaker3: [00:54:58] Word, a graduate degree could light a fire under you, right? And get some real guidance from people who have done stuff like that before to really give you that support. So the Harpreet: [00:55:06] Main thing is, does it align with exactly Speaker3: [00:55:10] What you want to learn? Like I, I managed to get access to, you know, Earth observation data and expertize in robotics and things like that Harpreet: [00:55:19] That I wouldn't have had within my Speaker3: [00:55:22] Day job and within my just googling for data sets. So those are things that aligned with my background in robotics anyway. Harpreet: [00:55:29] So I needed that. And that was Speaker3: [00:55:31] The reason why I needed a master's program because those are things that you don't have access to. So when you're looking at a master's program, figure out what's your timeline? Do you need a two or three year graduate program or do you need something that's just going to teach you the fundamentals in a year, year and a half? A lot of British universities and European Harpreet: [00:55:47] Universities, they're moving Speaker3: [00:55:49] To a year program because you learn the fundamentals and the the rest of it. You guys are right, you learn way more Harpreet: [00:55:53] On the job, right? Speaker3: [00:55:55] So figure out what are those research groups that you're working with? Is that going to add value to [00:56:00] where you want to focus as a data scientist? Harpreet: [00:56:02] There's data science problems all over the place. Speaker3: [00:56:04] Focus yourself and figure out Harpreet: [00:56:05] Exactly which bits Speaker3: [00:56:07] Are going to add value to you Harpreet: [00:56:09] And really consider the cost. Speaker3: [00:56:11] Really, consider the time because graduate degree is like, I know the status of my bank account. What what master's degree did to that over the span of just a year, right? So think really hard about how much time and money you're investing. Speaker5: [00:56:25] I've always come to that. The that same conclusion that I don't know if the opportunity cost Harpreet: [00:56:33] Outweighs the tuition for the vice Speaker5: [00:56:37] Versa, the opportunity cost would be greater than the tuition in the time and energy would go into a grad program. So I have not Harpreet: [00:56:43] Gone, but I'm Speaker5: [00:56:45] Starting to reconsider that Harpreet: [00:56:47] And I just want to get thoughts Speaker3: [00:56:50] On that. Yeah, I mean, like for me on that front, I like I took a very calculated approach to this. I listed like some 80 or 90 universities across the world, some 140 odd courses and really figured out which course is going to give me the information that I couldn't learn from myself in the shortest period of time for the lowest amount of money that I can give it right. I can do is like just gain as much out of it. I've seen a bunch of comments around this on like the Facebook subtle engineering traits page and a couple of people on there giving some great advice on Harpreet: [00:57:22] Make sure that let's say you're going Speaker3: [00:57:23] To pay $40000 dollars for a master's program. I don't know what the cost is in the U.S., but let's say it's 40000, right? Harpreet: [00:57:30] Try to make Speaker3: [00:57:31] Sure that your career growth potential within a couple of years of finishing that master's, your salary gap should increase by the cost that you should be able to recover that cost. It's an investment in you to treat it as an investment, right? Harpreet: [00:57:43] Yeah, absolutely. Knowledge bombs. Thank you so much, coach. So let's go to mark after Mark J.R. and Matt. Harpreet: [00:57:49] Great discussion kicked off by Magat. Thank you so Harpreet: [00:57:51] Much and then carried on by Matt. I appreciate that. But Mark, go for it. Speaker5: [00:57:55] I was going to say I'll let others speak for us. I talked a lot, so we'll hear from others have [00:58:00] spoken. Harpreet: [00:58:00] Yeah, let's go to J.R. and Matt and then back to Mark. So Jay, go for it. Speaker5: [00:58:04] Yeah. So my background is in business. I do not have see as background data science background, Harpreet: [00:58:10] But I went to grad Speaker5: [00:58:12] School, I took business classes Harpreet: [00:58:14] And so forth. And I mean, I was Speaker5: [00:58:16] Thinking I had to go to grad school for data science, but thinking about how much it cost, it's a lot. So I Harpreet: [00:58:21] Want to make sure that whatever Speaker5: [00:58:22] Money I'm investing in grad school for data science, it should Harpreet: [00:58:26] It should mean Speaker5: [00:58:27] Something to Harpreet: [00:58:27] Me. So before doing that, Speaker5: [00:58:30] I I took classes in Udacity Harpreet: [00:58:33] And Coursera and all that classes and to Speaker5: [00:58:35] See whether it's something I want because number one, I don't have a Python background. So I took classes there and I taught myself how to program Python R and all that stuff. Harpreet: [00:58:45] So take those Speaker5: [00:58:46] Classes because investment that's slightly smaller than grad school, I feel. So get a feel for it. And you know and see if this is something you really Harpreet: [00:58:54] Want to do and do one course Speaker5: [00:58:55] And do another one, then another one. And just see, I mean, I mean, Harpreet: [00:59:00] And I can the reason why Speaker5: [00:59:01] I'm doing Data is Harpreet: [00:59:02] Because in Speaker5: [00:59:03] Any organization that you go to Data, it's like like the thing in every organization and you have to know. So that's why I want to know how how Python works, how our works and how Harpreet: [00:59:13] Data works and all the situation. Speaker5: [00:59:15] And I find taking classes and all this different MOOCs super helpful. And then you can think about whether you really want to do a master's or a PhD. But I find this MOOCs are super, super awesome because in short time, you can get a lot of learning done pretty quick and fast, and then you can still get a job and still get the same pay, whether your masters or not. But yeah, having a master's Harpreet: [00:59:43] Program, you follow through some courses step by step and Speaker5: [00:59:47] Do the research part of it Harpreet: [00:59:48] Or a capstone project. Speaker5: [00:59:50] Either way, but that's my take. Harpreet: [00:59:53] Thank you very much. Matt, let's go to Matt and then after Harpreet: [00:59:56] You, we'll go to Mark. Yeah, I feel like [01:00:00] it's it's all Speaker5: [01:00:00] Just a big analytical problem, right Harpreet: [01:00:03] To to approach. Speaker5: [01:00:05] I try to think of it very pragmatic. So what? What are my goals? What am I trying to accomplish and what is it worth to try to get there faster if this path a Harpreet: [01:00:15] Masters is going to help me get there faster Speaker5: [01:00:18] Than maybe it's Harpreet: [01:00:19] Worth it? And so I Speaker5: [01:00:20] When I was thinking about master's programs, I did a lot Harpreet: [01:00:24] Of research. I came up with my Speaker5: [01:00:25] Goals, decided what it was that I wanted to achieve, and then I got into my Harpreet: [01:00:29] Top pick and I got to talking to two alumni. And one of the Speaker5: [01:00:33] Things that I kept going back to was make, man, this is expensive. Harpreet: [01:00:37] And the feedback I got from some folks Speaker5: [01:00:39] Was, Harpreet: [01:00:40] Look, if the cost Speaker5: [01:00:41] Is too much, either your goals are too low or you shouldn't be doing it, that's it. So either reevaluate your goals and make it make sense or just don't do it. Go a different route. Maybe it's not for you. And so I had to take a step back and realize a lot of the visualization I was doing. It's like, Yeah, sure, I had goals, but I was still picturing myself at the end of this program. One hundred and thirty grand Harpreet: [01:01:03] In debt without a job Speaker5: [01:01:05] Or an equivalent job to where I was, I wasn't picturing myself at that next level with what I could achieve by achieving this Harpreet: [01:01:11] Thing, right? Speaker5: [01:01:12] So building that into your process and having these goals and using it to help you get to somewhere else that it's got to make sense. Now, if there's like personal development, I can't help you there. If there's all these other softer reasons that's for you to decide. But from a financial perspective, you should be able to do the math to figure out what is your time worth. What is this investment really worth? And is it something that you want to pursue? Kind of to some of the earlier topics talking about what are what are programs looking for? They're going to be most protective Harpreet: [01:01:42] Of their asset, which is their alumni, Speaker5: [01:01:46] And then you will become their alumni by Harpreet: [01:01:48] Becoming one of their cohort. So the question Speaker5: [01:01:51] Was asked, like, what did you get from a program that you couldn't have got out in the real world? Well, it sounds cheesy, but the reality is the network, the people that I sat [01:02:00] with side by side from all over the world, from all different backgrounds, with anywhere from 10 years to two years of experience who shared their stories, shared their their experiences, added their two cents that are completely different from what anything I would have experienced elsewhere. Harpreet: [01:02:16] That's the value that was Speaker5: [01:02:17] The experience that I thought made it all worth it was you come away with all these new ideas, perspectives and connections, and then you step out and now you're part of this alumni network and there's tremendous value in that. So that's all stuff that you've got to take into the equation. But anyway, that's my two cents. Harpreet: [01:02:33] Thanks very much. Appreciate that. Let's go to let's go to mark. And then somebody was just talking. I don't know who that was, but let's hear from Mark. And then I guess after Mark, we'll move on. I got a question coming in from LinkedIn. Harpreet: [01:02:45] But whoever that was that Harpreet: [01:02:47] Wanted to say something. Just raise your hand and all Speaker5: [01:02:51] This building real quick on that the network component, the network is really huge. I've talked to here before about I tried starting a Harpreet: [01:02:58] Company and my two Speaker5: [01:02:59] Co-founders were my classmates from from grad school. And we liked working with each other so much or like for the next time we do something like we'll be co-founders again. So I wouldn't have got that without kind of having that networking component. But the kind of perspective I wanted to bring was my program was analytical because it's very research focused. But the reason I went to my program was to help me get into med school, Harpreet: [01:03:22] Not to Speaker5: [01:03:22] Go get a career in data Harpreet: [01:03:23] Science. It just so Speaker5: [01:03:25] Happened that my my master's was integral for me, breaking into data science. And the reason was it wasn't for my statistical skills or my technical skills. I built up domain knowledge that very few people had in health care, that a lot of people have it, but it's a harder to find skill than than others. And so when I was hired for my first Data science Harpreet: [01:03:48] Job, they're like, Speaker5: [01:03:49] Yeah, you kind of suck at technical stuff right now. You can train up on that. But like, you know how to work through kind of health care research in a way that that's a skill set that's highly valued. [01:04:00] So another component is like, All right. Beyond just like an analytics program, is there a domain that you're really passionate about? The has a analytics piece to it, and so I force my program to be analytics. I went to my director. I said, Hey, you know, I want to take every single stats class. Can we restructure the requirements I negotiated with my? My director essentially always negotiate. That's another Harpreet: [01:04:24] Component. But saying, like, Hey, Speaker5: [01:04:26] I want to take these courses in statistics, that's how it aligns with the program. And so I was able to take every single like simulation modeling and stats course I can get into, which ultimately end up helping made me realize, actually, I love data science through that. And the reason why I did is with my background in undergrad was qualitative research. So I wanted to learn quantitative research to mix methods and as a as a physician. And so it just just didn't end up that way, so I think something that's really useful is like actually getting domain and subject matter expertize and a subject can go a really long way to kind of get your foot in the door for your first Data role. Harpreet: [01:05:04] Thank you very much, Mark. There's a awesome comment from Monica Monica. Harpreet: [01:05:08] I like this comment. Go for it. Speaker5: [01:05:10] Yeah, for sure. I think that the biggest benefit, at least in my opinion, to graduate college is really to learn how to learn and like soft skills like time management. Because you have all of these courses, you're trying to get assignments on time. Harpreet: [01:05:24] You have a job. You're just figuring Speaker5: [01:05:26] Out how to be an adult. So there's just so much going on at the same time. So those soft skills I didn't have Harpreet: [01:05:35] Data science wasn't Speaker5: [01:05:35] A degree for me. So I have a business as well. And something that Harpreet: [01:05:42] Directly, I guess connects Speaker5: [01:05:44] And helps me is like understanding how to gather and adhere to business requirements. Because your assignments like they're given, you're given instructions and how well you meet those instructions, the better your grades get. So I think that's a direct [01:06:00] tie, even if you don't have a science degree. Harpreet: [01:06:04] But if I might just interject Speaker3: [01:06:05] There for a second. Monica, does that say something about the way in which our like school systems are preparing people for adulting? But this whole concept of Oh, we learn to be an adult at the university or at the graduate Harpreet: [01:06:18] Level, right? These are skills that Speaker3: [01:06:20] You don't need a graduate degrees into understand right now. I don't know much about the American or the Canadian systems and education. I've started developing a hypothesis of my own on how Australian education might need to evolve to meet some of those skills and talents. I'd love to hear more about like what you think on how that path looks like in Canada, for example. Speaker5: [01:06:40] Yeah, that's a great point. There's a lot of topics of discussion around courses that are given to high schoolers where like cooking classes or how to how to write a checkbook book, and nobody has checkbooks anymore. But that was something that was discussed when I was there, but that was never that was never offered. So people, they didn't know how to manage their finances. So then when you had to pay for all of Harpreet: [01:07:05] Your graduate Speaker5: [01:07:06] Degrees afterwards, you're like, Oh, I don't know, let's just get credit cards and figure things out later. So there's definitely room for opportunities to enhance the classes given to high schoolers and even middle schoolers. Harpreet: [01:07:21] I think we should push classes like, Harpreet: [01:07:24] For example, pre-calculus trigonometry calculus, push that to college, replace that with logic, business math, you know, simple classes, a bunch of taxes, things like that, right? Because not all high schools, like the vast majority of people who graduated from high school don't actually go to Harpreet: [01:07:43] University, right? Harpreet: [01:07:45] It might be. We're biased because I'm pretty sure all of us here have some form of higher education, but we just kind of see what we see. But all that, all that shit should just be pushed for. You know, if you're going to study math or engineering or computer science, just push it to there Harpreet: [01:07:59] Because they [01:08:00] don't really Harpreet: [01:08:01] Serve. I mean, definitely teach geometry. I think geometry should be taught in high school. Just there's a certain way of thinking, but yeah, it looks like you want to say something. Go for it. Speaker3: [01:08:12] I mean, I kind of agree and disagree with you on a point. I think we need to start embracing the idea of more open style learning Harpreet: [01:08:20] As opposed to necessarily Speaker3: [01:08:21] This cookie cutter thing where everyone learns the same thing because it's not going to make sense. Like for me, I want after high school, I wanted to go into robotics now for something like calculus. It's not something that you'd really master in, like three or six months, right? Harpreet: [01:08:35] Like the fact that I Speaker3: [01:08:36] Had four years in high school where we were doing calculus at varying degrees of intensity, right? You need that time of the sheer practice with mathematics to to get to any kind of usable level, right? And now I totally agree we are biased Harpreet: [01:08:49] In this room that all of us Speaker3: [01:08:50] Kind of have Harpreet: [01:08:51] That background have benefited Speaker3: [01:08:52] From it and other people wouldn't benefit from it at all. So I think we need to start thinking Harpreet: [01:08:57] More flexibly because the fact that we've Speaker3: [01:08:59] Stayed in this kind of rigid teaching structure has effectively meant that the gap between the skills that industry needs and the skills the school prepares you for, that gap is growing a lot wider, and that's forcing this need for tertiary education and then master's and graduate studies Harpreet: [01:09:16] And, you know, for entry Speaker3: [01:09:17] Level jobs. So I think that gap needs we need to address that gap sooner rather than later. Otherwise, we're just waiting till people are 35 before they can be, you know, horrible at all. And I think that's just not tenable for a society. So I don't know. Maybe that's just my philosophical stance. Speaker5: [01:09:34] Go for it, Greg. Yeah, I was going to say, I think if teaching systems, our school systems focus more on how to train people on how to fix Harpreet: [01:09:45] Problems, how to Speaker5: [01:09:46] Problem solve how to use. Your soft skills to attack different use cases, problem cases. It would be a far more powerful thing you wouldn't hear professionals [01:10:00] today hear about, Hey, why do I need to do to learn how to enter the health care industry? You wouldn't hear that because in school I've been taught to approach different problems, different ways or figure out what tools best fit, a specific problem or a list of set of problems. So and with that problem solving skills, I feel like is is lacking, come up with a list of use cases and walk me through how and why you would leverage this strategy versus the other one and show me how to surface know which strategy is best and why. When it comes to certain problems, at the end of the day, you know, learning technical stuff or special skills at school will not give you everything until you get on board with a company and learn what makes them tick Harpreet: [01:10:58] Or what their Speaker5: [01:10:59] Business model is. That's what makes you an expert, right? So when you have a degree, you're only telling that employer that you have the capacity to learn their business model and bring up and execute what they want you to execute. Harpreet: [01:11:15] So focusing on problem solving Speaker5: [01:11:17] Skills, which is, by the way, including how to work with people, how to empathize, how to Harpreet: [01:11:26] Explore, test Speaker5: [01:11:28] To me is a far more powerful thing in which, to your point, Harp you were saying, why don't we push these things math and stuff like that in college and teach kids Harpreet: [01:11:41] The different things? So I Speaker5: [01:11:43] Would Harpreet: [01:11:43] Value teaching problem Speaker5: [01:11:45] Solving skills in high school a lot and even in, you know, bachelor's and master's and things like that as well. Harpreet: [01:11:53] So it's something that is Speaker5: [01:11:55] Beneficial everywhere you go, right? Whatever industry you [01:12:00] cross over from or go to problem solving Harpreet: [01:12:04] Skills is huge. Love that to, you know, to quote Harpreet: [01:12:08] Big Sean, why don't schools teach more mathematics, less trigonometry, more about taxes, Tom, go for it. So, Tom, we can't hear you. You are unmuted, but for some reason we are unable to hear you, Tom. In the meantime, anybody want to speak on on this topic? Definitely. Let me know, Tom, you can try and interject and see what happens. It doesn't look like anything's going on there. Just hold your peace until we get past this next question. This will be our last last question here. Coming in from Dennis on LinkedIn, Dennis wants to know How do you find Data for portfolio projects that aren't as clean as Kaggle, but raw enough to showcase skills? I have a difficulty finding Data to begin with censorship. Government websites have been a nightmare. Harpreet: [01:13:01] Our mandate is everywhere, man like Harpreet: [01:13:03] It is literally everywhere. You've got it on your Spotify music listening history. You've got it on your wearable devices. My car collects data on Harpreet: [01:13:12] My driving, which I have to Harpreet: [01:13:15] Sign out of the app on my wife's phone because she's my driving scores. You'll be quite upset. Data is everywhere, man. Like literally everywhere. I see there's a bunch Harpreet: [01:13:23] Of stuff that Rodney. Harpreet: [01:13:25] By the way, Rodney, thanks for all the wonderful comments and LinkedIn. You know Harpreet: [01:13:29] You can. You can find Data anywhere you just have to think. I think instead of, Harpreet: [01:13:32] You have to flip the question around a little bit instead of thinking about how you find Data. Just think about what questions should I try to solve or what questions should I try Harpreet: [01:13:41] To make progress Harpreet: [01:13:42] On? What questions should I try to answer? Once we clearly define the question, Harpreet: [01:13:46] Then from there I think Harpreet: [01:13:48] The finding of Data, it's an easy next step. But I'd love to hear anybody else's perspective on this. If there are other perspectives. Yeah, everybody's saying Data is everywhere. Get an API. [01:14:00] Find it. Speaker3: [01:14:02] I yeah, I guess. Like, You're right. Find a problem and create your own data set. It's very easy to go and find a well curated dataset, and that's great if you want to learn about the models. But if you want to learn about deployment, because that solves that first step for you. But if you want to learn about the data, go build a data set and then you will build it wrong. I guarantee it and then it will break, and it will make your life hell and miserable. And then you will figure out why you shouldn't have created it the way you did. And then you'll learn more about how to create the next dataset and then you'll just improve that way. So it depends which part of the pipeline you want to learn about if you want to learn about models or for the off the shelf, if you want to learn about data cleaning and data structuring. Don't worry about the models, find a problem and create your own data set. Harpreet: [01:14:43] For example, let's say you're interested in figuring out the average time it takes between cars passing by the window of your house, right? Great. You can collect that data manually, or you can make it into a computer Harpreet: [01:14:54] Vision problem, right? Harpreet: [01:14:55] Like, I'm just going to set up a camera and just needs machine learning just to count cars as they pass by. And then I'll take that data and then just fit. I don't know what it's an exponential Harpreet: [01:15:05] Distribution Harpreet: [01:15:06] For waiting time and, Harpreet: [01:15:07] You know, predict the Harpreet: [01:15:08] Number of cars that pass by my window on a given day. You just think of interesting questions and, you know, find interesting ways to solve them, I guess. Harpreet: [01:15:17] Hopefully, that didn't extend. Yeah, yes. Harpreet: [01:15:19] You're back, Speaker3: [01:15:21] You're back at the big boy. There you go. Harpreet: [01:15:24] So let's let's integrate brilliance, Monica Speaker3: [01:15:28] In the chat and enter statements. Learn how to learn. Greg talking about problem solving. So, Huston, you are going to find endless things. You have to work on your whole life. See this. I'm still dealing with personal crap of my own. Ok, so guys, just to caution, we don't need to learn everything in school and that's a faulty mentality. Oh, I need to learn this. Go back to school. I promise you, I am not going back to school. [01:16:00] Dr. AIs is not going back to school. If I can't learn on my own now, shame on me. And really, Monica's right. We've got to learn to learn. So this is a big point. Develop your own learning plan and make sure you have a learning matrix Harpreet: [01:16:17] And don't think of just learning by topic. Speaker3: [01:16:21] Also, think of the other dimension Harpreet: [01:16:23] How deep have I gone Speaker3: [01:16:24] On each topic? And don't worry about going super deep on every topic. Some of it may be just I need to know how to use this. Harpreet: [01:16:33] Some of it may be. I need to know Speaker3: [01:16:35] How to derive this from scratch mathematically and then code it without any libraries. And then there's everything in between. But you have to you have to be the master of your own learning plan because there's too much to learn everything. No one person can even keep up with everything going on in Data science, but you can learn to a degree where you know how to learn very rapidly what you need to learn because you've focused on key concepts. Now what are the key concepts? You'll figure it out costed that that's part of mastering your own learning plan to you. Harpreet: [01:17:14] For me, Speaker3: [01:17:15] My key concepts. May be a little different than yours. Some of it's going to be guided by passions, some of it's going to be guided by needs in your role. But I hope that helps. I think mastering your own learning plan and not expecting school to fix all your problems, those are those are big ones. Harpreet: [01:17:33] I love that it reminds me of this quote. Like not I'm not going to quote it, but this general feeling that David Deutsch was talking about his book The Fabric of Reality. He's talking about how Harpreet: [01:17:43] Back in the days when we Harpreet: [01:17:44] Didn't have much knowledge as possible Harpreet: [01:17:46] For a learned person to Harpreet: [01:17:48] Know everything there was to know, that's not possible anymore. But it is possible to understand the explanations of things. I think that's what's Harpreet: [01:17:58] Really foundational is just being [01:18:00] able Harpreet: [01:18:00] To understand explanations. Then, you know, you're in a good position, by the way. Shout out Ben Taylor, Harpreet: [01:18:05] You're in the building. Harpreet: [01:18:06] See here, man. Harpreet: [01:18:08] Somebody unmute it. Go for it. Harpreet: [01:18:10] If now we can begin to wrap it up. Speaker5: [01:18:12] Isn't that what machine learning does? Learning to learn? Yeah, it's Harpreet: [01:18:17] It's something that we teach machines. Speaker5: [01:18:21] Maybe we can teach ourselves as well. It's quite funny. Harpreet: [01:18:25] Yeah, I feel every man. Yes, that is what machines do. They learn how to learn. Speaker3: [01:18:31] If I may Harp. Yes, that's a good point. I literally when I was learning reinforcement learning, I was so humbled by it. I don't mean by the complication of learning it. Harpreet: [01:18:42] But then if Speaker3: [01:18:43] I when I step back and abstracted what the agent was doing, I'm going to change the language a little bit. It came up with a really bad plan, but it operated on that plan and then it stopped and it looked back and said, Hey, how can I improve on that? So it it used some experience to refine the plan, operate some more, refines the plan. I've been trying to be very cognizant to do that transformative. That's why I bring up the learning plan, too. But also the life plan. I love the Gallup strength finder, but I think it missed the mark a little bit. I think it's also good to say, what do I suck at? That's really holding me back and how can I fix that? That's what Tom AIs has to do, and I learned that from reinforcement learning. It's not like we don't do it because we we realize, Oh, I better fix that, but to be more cognizant of it, that's what really helped. Speaker5: [01:19:40] Thank you very much. Harpreet: [01:19:42] I guess we'll go ahead and end it there. Thank you guys. Harpreet: [01:19:43] So much for joining. Don't forget to tune in to the episode that I released today, but I see Greg unmetered. So sorry, Greg, if there's a follow up, Harpreet: [01:19:51] I didn't mean to cut you off. No, I get. Harpreet: [01:19:53] Hopefully, you guys get to tune into the episode released earlier today with Pradeep Senga. It was cool talking to talking to him. Harpreet: [01:19:58] Definitely. Check out that Harpreet: [01:19:59] Episode. [01:20:00] Also, big shout out to the Narrative Science podcast Cassidy Shields had me on his show earlier today. Really enjoy being on that on that podcast. Also, Monica, you got your present. No, you're hosting something. Shut up. Shut that out. Speaker5: [01:20:16] Yeah, yeah. Actually, it was yesterday for Data Science Go Connect. However, I will also be at Go Virtual next month on August 15th. So if you guys want to go there, Harpreet: [01:20:30] I'll be there. Yes. Harpreet: [01:20:32] October, October 15th. I messed that up earlier today too. Harpreet: [01:20:35] I called October August earlier today. I don't know what that Harpreet: [01:20:38] Was, what that's about. Hopefully, you guys sign up for the dedicated conference that's popping off in a couple of weeks. I'll be presenting there. Also be presenting at the Machine Learning Conference on October 15th on behalf of Comet, talking about MLPs and things of that nature. Antonio, the CSA community announcement as well. Let it go, man. Let us know. Speaker5: [01:20:59] Oh no, I was just saying October, but October. Harpreet: [01:21:02] Yeah, yeah. So yeah, guys, don't forget to also tune in on September 22nd, which is this coming Wednesday? Harpreet: [01:21:11] Me and my good Harpreet: [01:21:12] Friend Sadie St. Lawrence will be live on Instagram. So if you are not already following me on Instagram, Data AIs Harp is the handle and the Save St. Lawrence just her name spelt out. Sadie Stella W R.N.C St. Lawrence will be live on Instagram Harpreet: [01:21:28] Will be a lot of fun. Harpreet: [01:21:29] Sadie's, the founder of Women and Data, have been having massive impact, so it's going to be a great conversation. Hopefully you guys could join us. I'll be speaking to Mr. Brant on October 2nd. That will be live on LinkedIn Brant. Thank you so much for sending the book. I've got to take a picture and Harpreet: [01:21:47] Post it, but I appreciate you Harpreet: [01:21:48] Sending the book out. So we'll be talking on October 2nd, a couple of other live events, so we'll be live with Brant Sykes. I'll be live with Natalie Nixon. We're going to be talking about the creativity leap that [01:22:00] is also in October. The date escapes me and then also Britney, though we'll Harpreet: [01:22:05] Be talking about her book bigger Harpreet: [01:22:07] Than leadership. So those are going to be some live sessions on the podcast and then interviewing a bunch of awesome friends coming up the next few weeks. I'm interviewing Joe Rees. Joe, what up, Eliana? You know, Eliana also Harpreet: [01:22:25] Just many, many people, many friends. Speaker5: [01:22:28] Is it is it too late as Ben for his fishing trip? I was that. Well, the good one witness that you brought to the table and what was the result? So it went really, really well and a little bad, but mostly good. Good. Good. Yeah, the weather couldn't have been better. I think our group got close to one hundred fish. Wow. I could not be happier with the shots that we got. There's so many variables with this trip, right? Like, so many things could have gone wrong. Harpreet: [01:23:00] But yeah, Speaker5: [01:23:00] I think the edit will be public in two to three weeks and then the Data set will also be public as well. So, yeah, I'll I'll do a deeper dove, but I'd love to share some of the footage with Harpreet: [01:23:12] The crew in the coming weeks Speaker5: [01:23:13] Because it's it's jaw dropping. So I'm actually going there right now. I'm in Wyoming again. Oh, nice to wrap up the Data set. Harpreet: [01:23:24] I'm looking forward to that man. I'm looking forward to this deep dove. I could learn a lot from that hundred fish. That's a lot of fish sticks, man. Speaker3: [01:23:30] And I'm really missing Wyoming, buddy. Speaker5: [01:23:33] Yeah, I'm laughing to myself because we're talking about scraping Data, doing projects, Harpreet: [01:23:38] And I was chuckling because I'm like, Well, I'm going to get Speaker5: [01:23:40] A Data set where I have to pack a gun because I'm going into bear country. So I'm like, Yeah, it'll be good. This river, I'm fishing is saying. It's quite an interesting use case, so looking forward to hearing more on that. You know, talk about, you know, to me, it's kind of like, look into nature and come up [01:24:00] with some innovative ideas. So it's quite quite quite cool. Harpreet: [01:24:03] So looking forward to hearing more. Speaker5: [01:24:04] Yeah, thanks for asking about it. Harpreet: [01:24:06] All right, guys. Go ahead and wrap it up. Thank you so much for joining me. Shared again to tune into the podcast. Join me on Sunday, guys. I know, you know, J.R. It's been a while since he came on a Sunday. So come, come back. We miss you on Sundays. So you guys remember a comment off Harris on Sunday at 11:00 a.m. Central Standard Time. Be sure to come on that it'll be live here on LinkedIn as well. Be sure to check out these cool episodes of the podcast that are coming up and have been released. And as usual, my friends remember you've got one life on this planet. Why not try be some big cheers, everyone.