Harpreet: [00:00:09] What's up, everybody, welcome, welcome to the comet ML officers powered by the artist and Data science. I am your host, Harpreet Sahota. I'm super excited to have all these guys here. Harpreet: [00:00:18] Those of you tuning in right now, Harpreet: [00:00:20] Whether you're on LinkedIn, whether on YouTube, whether you are on Twitch, you are more than welcome to join us here. I'm going to drop a couple of links right there Harpreet: [00:00:28] In the comment Harpreet: [00:00:30] Section that you can click on to join us right here in the room. Or you can just leave your comments if you're shy. You can leave your comments right there in the Harpreet: [00:00:37] Chat itself, Harpreet: [00:00:39] And I'd be happy to answer your questions as they arise. Harpreet: [00:00:43] My friends, super Harpreet: [00:00:44] Excited to have all of you guys here. Hopefully you got a chance to tune in. To the episode I released on Friday with Dennis Will, he is a Data engineer based out of Berlin. We had a great chat. I know Dennis through through LinkedIn. We connected that way. I was sponsoring his LinkedIn page for a while because he's putting out such great content. Harpreet: [00:01:06] I always get to Harpreet: [00:01:07] Give back to creators. Man was all about super excited to have all the guys here. You know, we'll get into your questions as they arise, Harpreet: [00:01:14] But I want to. Harpreet: [00:01:15] I want to talk about something real quick. I want to talk about what makes Data science so unique, right? And I think there's three key reasons why Data science is so unique. One, it is a meta skill, too. It is Harpreet: [00:01:29] Permissionless and three, you Harpreet: [00:01:32] Can create Harpreet: [00:01:32] Your own Harpreet: [00:01:33] Experience. All right, let's dig into this a little bit more. First of all, how is Data science a meta skill? You're probably wondering what the hell a meta skill is, so mezcal Harpreet: [00:01:44] Is essentially just a higher Harpreet: [00:01:46] Order skill that enables and empowers other skills to happen. It's the foundation on which you're able to Harpreet: [00:01:52] Engage with new Harpreet: [00:01:54] Skills Harpreet: [00:01:54] And new knowledge Harpreet: [00:01:56] And technology more effectively. So in the case of Data, science is a [00:02:00] combination of several different knowledge bases and skill sets, critical thinking, Harpreet: [00:02:05] Problem solving, programing engineer, Harpreet: [00:02:08] Engineering, math statistics, business acumen communication, project Harpreet: [00:02:11] Management, just to name Harpreet: [00:02:13] A few. So a lot of university programs and Harpreet: [00:02:15] Boot camps out there, I've Harpreet: [00:02:17] Noticed, tend to focus primarily on acquiring the technical knowledge and teaching Harpreet: [00:02:21] You individual tools Harpreet: [00:02:22] Of the trade. It's because the rest of those skills that I just mentioned, they're very difficult to train and teach. Harpreet: [00:02:30] And those Harpreet: [00:02:30] Programs certainly, you know, don't teach you how to put all Harpreet: [00:02:32] Of that together Harpreet: [00:02:34] To make you into a cohesive candidate, a cohesive, robust Data science, so to speak. Harpreet: [00:02:41] The second thing I want to talk about is this concept of Harpreet: [00:02:43] Permission listeners Harpreet: [00:02:45] Permission less, Harpreet: [00:02:46] Right? So if we start by contrasting Data science to some other Harpreet: [00:02:50] Professions out there, especially Harpreet: [00:02:51] Those careers where you're legally required to have some specific certification or degree Harpreet: [00:02:56] To do the job, to become an Harpreet: [00:02:58] Accountant, Harpreet: [00:02:59] You have to enroll in the Harpreet: [00:03:00] Cpa professional Harpreet: [00:03:01] Education program, complete Harpreet: [00:03:03] 30 months of relevant accounting experience and finish four education Harpreet: [00:03:06] Modules during full Harpreet: [00:03:08] Time work experience. And I'm sure it's a lot more complicated than that right to become an actuary. Even after you complete a degree in math or stats, you need to pass a battery of exams and additional coursework that's required by these governing societies. And that's just a couple of these examples. There's many, many, countless career paths where you have to have permission by some governing body to be rightfully call yourself on Data crown, but we don't have that in Data science. That's because in Data science, you can create Harpreet: [00:03:40] Your own experience. Harpreet: [00:03:42] You just need a lot of creativity and ingenuity to make it happen. And the way you can prove that you have what it takes to do the work of a data scientist is by having a portfolio of amazing projects, right? So in those fields that we just discussed, the concept of having personal projects to demonstrate [00:04:00] your understanding and command doesn't really exist. It's not like a accountant can go do a personal accounting project and say, Hey, Harpreet: [00:04:06] I'm going to be Harpreet: [00:04:07] An accountant now, right? It's not like an actuary can do an actuarial project and demonstrate that Harpreet: [00:04:13] They have the the skill to Harpreet: [00:04:14] Do the job right. But in Data science, you can create your own experience, right? You know that anybody, Harpreet: [00:04:22] Anybody can become a data scientist. Harpreet: [00:04:24] You don't need anyone's Harpreet: [00:04:25] Permission, right? Harpreet: [00:04:26] Anybody can get set up with Python or Java. Anyone can download vs code or whatever ID of choice to and start writing code. Anyone can set up a SQL database on their local machine and start writing code, so it is entirely permissionless to break into data science like the barriers to entry are pretty much nonexistent. Obviously, you have to go through the interviewing process and they're tough, but they're tough for a reason. But that being said, I just want to go off on a little bit of a, you know, rant and tirade just to get the conversation flowing. I'm excited to be here with you guys to see a bunch of you tuning in on LinkedIn. Feel free to join us on LinkedIn. I'm going to be dropping right here, Harpreet: [00:05:10] The link for you to Harpreet: [00:05:11] Come and join us in the room. Please do. Harpreet: [00:05:13] And please come hang out. Harpreet: [00:05:15] So far we got married and we got Leah. Leah, good to see you. Happy to have all the guys here. So let's go ahead and just jump into questions. Merryn or Leah, if either if you have a question, let's let's go ahead and get started also. Marianne has a very cute, cute dog that as well. Harpreet: [00:05:30] So, so anybody Harpreet: [00:05:31] Have a question? I know Marion, you had a question before we we started broadcasting, so definitely feel free to ask that and let's dig into it. Marion: [00:05:40] Uh, yeah, my question is I sort of have. Harpreet: [00:05:45] Council dancers myself, I had the Marion: [00:05:48] Experience with the company into doing three weeks ago, I think I did very well on the first dance with the take home assignment. Uh, but if [00:06:00] they email that unfortunately were not moving with the application for the second round and actually I had requested a feedback from the hiring manager who was very kind to me. Harpreet: [00:06:14] And the next Marion: [00:06:15] Day, after I request the feedback, we get a, I think, 45 minutes chat and he started. Harpreet: [00:06:23] He was Marion: [00:06:25] Impressed with the submission, but there were Harpreet: [00:06:27] Just too many applicants. Marion: [00:06:30] The number that he gave me was really something mindboggling. He said that they had nine hundred applicants for the position of that scientist. And to me, I was thinking something like 200, maybe. Uh, so you had a couple of constructive points about my submission, but at the end of the day, he said there were just too many applicants and a bunch of them came up a little bit better than you. And I said, That's fine, but no problem. I know that the competition is very strong and there are people with more experience than me. Surprisingly, couple of days ago, three weeks after our session at the same person reached out to me via email and said, Hey, listen, would you be open to talk again? We had a call, and he said, I'm still interested in the position. How about we bring you for a second round? All sorts of. Yeah, of course I said yes, but also that parks and bells started ringing in my head because through the past it's not like one week and Harpreet: [00:07:50] It's all of a sudden Marion: [00:07:52] Decide to. To bring me in. And that's what what made in Harpreet: [00:07:59] The [00:08:00] Marion: [00:08:00] Home again. He said, well, listen, we had a very good session during our feedback call and you followed up with something that we have done looked at our blog page had done this, suggested some, some different approaches. I was impressed with that. Also, I looked at your project on Harpreet: [00:08:19] Github, and I also Marion: [00:08:21] Think that that's why I decided to call you again. So that's a valid answer, but that's probably part of the answer because I have sent him my links to the projects much earlier that I talked to feedback. Coal, though, believe he will lead it into the city and looking at byproducts to Harpreet: [00:08:44] Kill the time Marion: [00:08:46] And be impressed. The moment when he saw them, so my my thinking is. They had candidates that they brought second down. They didn't like anybody. That's one possibility or the person they like that said they Harpreet: [00:09:04] Got maybe five six two second down. Marion: [00:09:07] They didn't like the five of them like one of them. Made him an offer, made them an offer, and the person refused. So that is not thought. That's sort of the scenario that is going in my head. Yeah, I agreed. But the fact that he didn't leave any door open oxygen for me Harpreet: [00:09:31] Right away Marion: [00:09:32] The major. Uh, makes me a little bit hesitant. I'll continue with them, but I don't know what to make of it. And because. Because this is my first experience, something like that has happened to me. Yeah, I have a lot of experience in different Data. I don't know what it means in the realm of Data science. And so if you [00:10:00] guys have experience like that, so can sort of give some sort of Harpreet: [00:10:06] What is your what are your Marion: [00:10:07] Thoughts on that? Harpreet: [00:10:09] Pretty much the exact same thoughts as you have like, I mean, if you're in one of those bigger markets, I think you're in California. So naturally, you're going to have more competition for jobs just because there's more people there. Harpreet: [00:10:21] Right? Harpreet: [00:10:21] So that's just a natural byproduct of that. So nine hundred people for a job Harpreet: [00:10:26] Application in, you know, Harpreet: [00:10:28] California, I would say, definitely probably realistic. And you're right, that's pretty much what happened. They probably had a they probably shortlisted a few candidates, right? And from Harpreet: [00:10:41] That shortlist, a Harpreet: [00:10:42] Few candidates that went on to second round, third round or Harpreet: [00:10:44] Whatever, maybe their top one or top two picks, Harpreet: [00:10:47] You know, maybe one of them just flat out refused the offer. They got another offer had competing offers, so on, so forth. And that's likely what happened, right? You know, the top candidate for sure. Harpreet: [00:10:59] Didn't accept the offer for whatever reason. Harpreet: [00:11:01] It could have been that the salary was too low, could have been that they had a better competing offer. Harpreet: [00:11:05] Whatever the reason is, and now Harpreet: [00:11:07] They're just going back to their pool of candidates, kind of like their reserved candidates. So out of that 900. A year in the short Harpreet: [00:11:13] List of people, right? Harpreet: [00:11:16] So that's good. Like, they're still considering you as a candidate that still gets signal. I wouldn't take any red flag from it. I mean, that's just what happens in companies, right? Harpreet: [00:11:25] People interview a ton of people. People will try to. Harpreet: [00:11:30] At the end of the day, they're making the best choice for themselves as a company, right? And at that stage, with the candidates they had in the pool at that time, maybe some of the candidates were a better Harpreet: [00:11:40] Fit to proceed along with the Harpreet: [00:11:42] Interview procedure, right? And, you know, for whatever reason, that didn't work out, and now they're going back to to some of the candidates that they had like, Harpreet: [00:11:53] You know, in Harpreet: [00:11:54] Reserves, so to speak, like. That the people on the bench, Marion: [00:11:58] I accept that it's Harpreet: [00:11:59] Just [00:12:00] to seems three weeks Marion: [00:12:01] Is a long period, but again, I draw on my experience in different fields electrical engineering. There are the internal processes needed for capital work and you probably have gone through all the possible candidates and. Or Data, you know, you take every step of Harpreet: [00:12:22] The job search process for Data, science is a bit bit longer and I can imagine this is kind of towards the tail end of summer here. People are on vacation, people are out of office, whatever. It's hard to coordinate schedules. If this is a company where a lot of the team is remote in different time zones like, you know, things like that, scheduling conflicts Harpreet: [00:12:41] Can can happen. Harpreet: [00:12:42] So, you know, they probably had like three or four candidates that they moved after the first round and interviewed three or four candidates could take three or four weeks, right? So it could take some time. So that's just, I think, standard. I mean, you might get lucky and get a quick, quick interview process, but typically, I think. Three weeks probably is about average. Should it be shorter? Harpreet: [00:13:11] Probably, but you know, OK, Marion: [00:13:12] I shouldn't take this as something like it's something unusual. Yeah, no. Why so late? Harpreet: [00:13:22] Yeah, exactly. Harpreet: [00:13:23] Yeah. Yeah. Marion: [00:13:23] Ok, that would be good. Thanks. That's study. Harpreet: [00:13:28] Yeah, no problem. Leah, how's it going? How's your weekend been? How's your week looking? Oh, coming up? Pretty good. Leah: [00:13:36] Pretty good. I'm happy I was able to pop in today. I've been meaning to for a while, so following through with my commitment to do that, so that's good. And Martin, thank you for kind of sharing about that and about kind of the questioning of, oh, you know, is this a red flag? Is this taking too long? I'm going through the interview process myself and doing it over at the summer [00:14:00] like it's I've noticed it's really slowed down. I recently relocated from the United States to to the U.K. I live in Bristol, England, Harpreet: [00:14:09] Now and I kind Leah: [00:14:10] Of had to Harpreet: [00:14:11] Learn the rhythm Leah: [00:14:13] Of like recruitment in a different country as well. So keeping in mind summer holidays and just a different environment, I really had to practice some resiliency and patience about putting applications out and having people get back in touch with me. And I think just by being responsive, you and yourself and the fact that you had that follow up call is awesome and you obviously made a good impression that that kind of triggered Harpreet: [00:14:43] The hiring Leah: [00:14:44] Manager to go and and to look at your portfolio. And again, that as you mentioned at the first about how having a portfolio is so important and how anyone can do it, like you don't have to have the gods of Data science bestow upon you the ability to do a portfolio. You can just do it for fun and share your best work with others as an example of what you can do. I wouldn't take it as a weird thing a red flag that they didn't look at it until a little bit later. I'm having to practice kind of some empathy about the hiring managers. They're people too, and they're busy and they're looking through. But I just kind of judge I judge it by like the interactions that I have with people. And if there's positive and I follow up and I'm putting in the energy to show them that I'm interested, I just kind of take that as I'm doing my best and they're doing their best. And as long as the interactions that I'm having with them are good, I try not to get too hung up on on the other stuff. So that's just my two cents in my experience. Marion: [00:15:44] Thanks. Thanks for the feedback. I have a quick follow up specific comment to that one. When that person told me, the current manager told me that they have hundred candidates for the position and right [00:16:00] the way the quote in my head came. Hey, listen, I'm transitioning from a different field in my resume. I have on the I have a data science book, six to nine months project pilot project and so on skills. But the thing that I'm studying is basically the puzzles Harpreet: [00:16:21] That I have found. And when I put an Marion: [00:16:24] Ad adding measure against nine candidate candidates, then they're just going to see what kind of projects I have done. But they're never going to click on the link and try to see what the product is. So how I know one of the answers how to overcome that is basically when you apply for something to reach out to the person who posted the project and say, Hey, listen, I applied for this position I can download. That is very similar to whatever the role you are looking for. You can check my projects. And so but even then and I have been doing it better for the industry, so to speak. But even then, when they're are so many candidates, I don't think they have the time to. Even if you have that, let's say I'm standing LinkedIn in mail to them. I don't think they have the time to basically go and pay Harpreet: [00:17:20] Attention that they had, Marion: [00:17:21] This Harpreet: [00:17:22] Person reached out Marion: [00:17:23] Separately, but I just cannot go and check this project. So there's a thing you're making now. How do you actually overcome that obstacle because. I don't see another way. Harpreet: [00:17:39] Well, you can get nine hundred resumes, it doesn't mean you're going to get nine hundred portfolio projects Harpreet: [00:17:44] To skim through. Right. Harpreet: [00:17:45] So by you having portfolio project that really honestly puts you in like the top five percent of candidates, most people applying for Data AIs roles and I've seen literally thousands of resumes. They don't have projects. They just, you know, they're not putting anything [00:18:00] on their portfolio so that that, you know, helps right there. And plus you can get 900 resumes. But out of those nine hundred resumes, like realistically going to shortlist 30 or 40 and which one which thirty or forty eight and short Harpreet: [00:18:13] List ones where the Harpreet: [00:18:17] Ones that are esthetically pleasing, let's say, put it that way, right? So resumes that look good on a first date, you know, kind of what's the word I'm looking for first like impression? So, yes, let's say for that. I don't think you're at all getting like, you know, they're not looking through every project, they're looking through the best candidates projects and even then, it's not one person that's looking at it. It's usually going to be members of the team that are going to be looking at it. Somebody saying sometimes tasks. So a couple of comments coming in on on LinkedIn here. Robert Robinson says Good, Celia Pope, Harpreet: [00:18:58] It's good to see the pope here as well. Harpreet: [00:19:00] Shout out to Asha, Harpreet: [00:19:01] Just joined in. Harpreet: [00:19:02] Vikas Singh says sometimes Harpreet: [00:19:03] Ml AML tasks Harpreet: [00:19:05] Feel like they're redundant. What do you say about that? Harpreet: [00:19:08] I say you're not Harpreet: [00:19:09] Doing enough machine learning or not doing enough interesting work. Harpreet: [00:19:13] That's what I would say about that. Harpreet: [00:19:15] Either that or you're just approaching every problem exactly the same and thinking Harpreet: [00:19:19] That just because you have Harpreet: [00:19:21] One way of solving a problem under your belt, that that is the way that all problems are solved. So that's what I would have to say about that, what what do you guys say, our ML task repetitive. I don't think the repetitive or redundant. What do you mean by redundant? Let's let's look up the word redundant, right? So being redundant, that means not or no longer needed or useful. Superfluous? All right. So sometimes ML tasks feel like they're not needed or useful or superfluous. Harpreet: [00:19:52] Yeah, OK, Harpreet: [00:19:52] I agree with you. That means Harpreet: [00:19:55] That. I mean, Harpreet: [00:19:57] If you're starting from the viewpoint that every single problem in the world [00:20:00] should be solved as an ml problem. Yeah, things are going to get redundant. That's not the case, right? Harpreet: [00:20:05] I mean, machine Harpreet: [00:20:05] Learning is not useful for every single problem. You know what I mean? And probably not even for every single quantitative problem, right? There are some problems that are just deterministic. You don't need machine learning to solve deterministic problems. You need machine learning to solve problems where there is inherent variability or probability involved. Harpreet: [00:20:29] Right. That's that's what I would say about that. Harpreet: [00:20:33] Uh, entry level Data science job often asked for skills that almost cover Data engineer analyst Emil Engineer How can one strategize job search do just do projects right? Like here's a three projects you should do as a data science aspirant. Listen to me. Three projects you should do and you'll cover everything there. The first project is do a Data engineering project and make it simple. Data is everywhere. Go to an API. I don't even care if it's the Weather Channel API. Harpreet: [00:20:59] Go to the Weather Channel API. Write a script that pulls weather every day at three hour Harpreet: [00:21:06] Intervals and then pull that data. Harpreet: [00:21:09] Do some stuff with Harpreet: [00:21:10] It, clean it and then dump it to a database. I don't even care if it's a local database or a cloud database, whatever. You've just built a ETL pipeline. All right, good. What do you do after that? Second project, if you do, is just an end to end machine learning project, right? And to end it could the deployment can be trivial. The deployment could be just serving Harpreet: [00:21:29] Predictions in a Harpreet: [00:21:30] Csv. The deployment could be stashing predictions to a database. The deployment could be a front end that is just deployed locally using stream lit or whatever Heroku app you want to get super sophisticated do on the cloud. But something that just Harpreet: [00:21:47] Takes Data, does Harpreet: [00:21:48] Transformations to it, has a model built, and then you're exposing that model to new Data. And that new data is going through the exact same Harpreet: [00:21:56] Pipelines and models predicting on Harpreet: [00:21:59] It and [00:22:00] predictions are stashed. Or, you know, doing one of those few things I just described, Harpreet: [00:22:06] Then you should also do a project Harpreet: [00:22:08] Here, says analyst. Yes, fine. Data come up with an interesting question, explored the Data do the work and see what interesting Harpreet: [00:22:17] Insights he find and communicate that right. Harpreet: [00:22:19] Like it's not hard to do. Each one of those projects will take, you know, more than a month and a half to do right. Each one of those projects fully Harpreet: [00:22:27] Thought out will take you a month and a half to do Harpreet: [00:22:30] It. Now, if you're a complete beginner, don't know anything about data science, then maybe you might take you three or four months to be one of those projects, which is fine. That's your learning curve. But this bullshit about like people complaining that these jobs? Yeah, OK, there are unrealistic job expectations out there. There are every fucking industry has that. Every industry has that. Do you just get Harpreet: [00:22:52] Over it like this? Harpreet: [00:22:54] Like this shit is just like feeling sorry for myself. Like, Oh man, like, what do I get? All these challenges, all these jobs out there, they want these candidates Harpreet: [00:23:00] To have well, then don't Harpreet: [00:23:01] Apply for those jobs, then dude. Like, just don't apply for those jobs. That's probably not a job that you should apply for. Harpreet: [00:23:07] There's hundreds of other Harpreet: [00:23:08] Jobs for every one ridiculous job that just sticks out in your mind. There are hundreds of other job opportunities Harpreet: [00:23:14] Out there that Harpreet: [00:23:15] Are more realistic, right? But this I can't catch a break. Attitude is it's just not. It's not going to get you anywhere in life. Rant is over. Just to stop Harpreet: [00:23:25] There and Harpreet: [00:23:26] And stop Harpreet: [00:23:27] Talking. See if anybody Harpreet: [00:23:28] Has Harpreet: [00:23:30] Any follow up on that because Harpreet: [00:23:32] I guess I can get worked up over shit like this man. Like, there's people like, I can't catch a break attitude. Like there's Harpreet: [00:23:38] This thing like like this Harpreet: [00:23:40] Mentality that is so hard to get off the starting block. You're not serving yourself well by thinking like that man. So how do you strategize Harpreet: [00:23:47] The job search? Build the skills. Apply for Harpreet: [00:23:50] Jobs. Reach out. Make content. Get noticed. Do whatever your competition is not doing. And just do what you can to stand out right? [00:24:00] Typekit, go Mary and go for it or let go from let's hear from let's hear from Leah first and then go to. Leah: [00:24:13] Yeah, yeah, like I've I've had to learn like I used to be such a Harpreet: [00:24:18] Stickler about, Oh, look at this Leah: [00:24:20] Job spec. It's got this laundry list. And if I didn't meet like 90 percent, I wouldn't even consider it. Now I'm kind of looking at, well, if I've got seventy five, if I have a reasonable answer to seventy five percent of the requirements, I'm going to go ahead and submit. Harpreet: [00:24:35] And if they don't, Leah: [00:24:36] If I don't meet the requirements, it'll get screened out. You will never get a job, you don't apply for it. So that's something that I've learned. And for the things, the jobs that I get really kind of excited about, I'll go a little extra little extra effort. And if I land like the first interview, I'll go ahead and try to Harpreet: [00:24:55] Do like a small Leah: [00:24:56] Project, typically like an exploratory data analysis Harpreet: [00:24:59] Project on Data for Leah: [00:25:01] That company. So I'll try to find publicly available data, whether it's Harpreet: [00:25:06] Like reviews, Leah: [00:25:07] Product reviews on Trustpilot or it's talked about on Twitter or Google. Google Play reviews. If it's an app Harpreet: [00:25:16] And just try to Leah: [00:25:17] Demonstrate I'm interested in your Harpreet: [00:25:19] Company, I can do stuff Leah: [00:25:20] With your Data and it might not even be a Harpreet: [00:25:23] Project that you're not going Leah: [00:25:24] To do a lot of machine learning on, like Trustpilot reviews where you've got like one hundred and fifty reviews, but you can do something. And so that's kind of the approach that I've that I've taken. And then to like, I don't if I meet seventy five percent of what they're asking for. I put my ring in the hat, my hat in the ring. You know what I mean? Harpreet: [00:25:42] And I would even say 75 percent, that's even like too much Harpreet: [00:25:45] I'd say if I make like if Harpreet: [00:25:47] I make 40 to 50 percent, I'll apply for it because the thing is like, you can learn stuff on the job, right? Like all these jobs require a certain fundamental baseline level of knowledge after which the things [00:26:00] are just just the they stack in compound, right? Obviously, like Harpreet: [00:26:04] You need to know how to code right. You need to be able to code, Harpreet: [00:26:07] But you don't need to know all the different, you know, tech stacks out there Harpreet: [00:26:12] Because that's just not Harpreet: [00:26:14] Optimizing welfare. Job search, you can learn all that stuff on the job, right? Yeah. Now, it's the excellent point. Thank you very much for that, Leah. Let's go to Mary and you're going to say something after that shout out to everybody else that joined us and he's in the building and dying is in the building. Matt Plaza in the building or in the building, you guys probably can't see them on the live stream. They have the cameras off, but they are here and they're hanging out. And if you want to hang out with us, you can as well. Harpreet: [00:26:40] And if you also Harpreet: [00:26:41] Want to hear me rant again, just just ask a question that triggers me and you will get a rant from me. Mary Ann go. Marion: [00:26:49] Yeah. Actually, there's is not much. But following up on the interview, I had the same comment in my head right away that no, I'm never waiting for seven to five Harpreet: [00:27:00] Percent, that if I Marion: [00:27:01] Meet 50, 50 percent of the requirements I'm applying, sometimes even. But the default is OK. The problem, I think that the person who has that question that is this Harpreet: [00:27:16] Sort of experience that I Marion: [00:27:17] Have when look for job descriptions to see a Harpreet: [00:27:20] Lot of cookie-cutter Marion: [00:27:21] Sort of descriptions, it's almost like it's copied and pasted from one company to the other. And that's the nature of Data stance. Harpreet: [00:27:30] I mean, it's kind of messy. Marion: [00:27:33] It's not well defined. People want to have all the bases covered, so they include everything. They include out. They put Python, but you possibly don't have that many people who Harpreet: [00:27:45] Know Box and Python Marion: [00:27:47] And have done work in both of them and Harpreet: [00:27:49] They're skilled in them. Marion: [00:27:51] So, so don't don't bother about not having a python if you have one of Harpreet: [00:27:58] Your Marion: [00:27:59] Good, so you'll [00:28:00] love the next thing they want you to have that. You'll learn it on The Jeffersons. So to that person, don't get discouraged about the cookie cutter sort of descriptions of the position. That's that's just the nature of it. Harpreet: [00:28:17] And as layers that. Marion: [00:28:20] Look at the descriptions that have some sort of unique being. And if you like that Harpreet: [00:28:25] Unique thing, then focused on that, but make Marion: [00:28:28] It happen, Steve, the company had something interesting that you can find about them and. What on a project that is Harpreet: [00:28:38] The Marion: [00:28:38] Position, so that that's the way to get noticed. Harpreet: [00:28:42] Yeah, man, thank you. And I just want to take a second here to read a couple of posts that my good friend just made just a couple of Harpreet: [00:28:48] Days ago to post Harpreet: [00:28:50] Back to back. And, you know, shout out to there's I think there's might be 20 people watching on LinkedIn live right now. Harpreet: [00:28:56] And if you Harpreet: [00:28:56] Are one of these people that are in positions to hire Data scientists, it is up to us to make the change right. But I'm just going to read out what Vin is saying here. Why can't companies find qualified data scientists? Their hiring process is built to eliminate them? It's not a secret. The major problems are obvious. One most senior data scientist wouldn't pass a junior Data science job interview because it has nothing to do with the actual job. Harpreet: [00:29:19] And this is so fucking true. Harpreet: [00:29:21] Two qualified candidates are bounced before they make it to the hiring manager because they don't keyword stuff their resume just the right way. Three job descriptions contain all the requirements for a data scientist, Data engineer and machine learning engineer, and they expect one person to cover three roles three plus rounds of interviews. Spend enough time looking for flaws, and you'll find them even if new hires require training and onboarding. So looking for the best or perfect candidate is unrealistic. I'm not sure how they expect to Harpreet: [00:29:50] Hire people when their Harpreet: [00:29:52] Entire process built to exclude people. It's not difficult to fix either, and then goes on and says that he's built rebuilt [00:30:00] hiring processes and got Harpreet: [00:30:01] Data scientists Harpreet: [00:30:02] Entire teams hired in six to 12 weeks. So, yeah, I absolutely agree with this. Yes, the hiring process is a bit broken, but that doesn't mean that you don't need to go and put in the work and Harpreet: [00:30:10] Gain the skills to make Harpreet: [00:30:12] Yourself look different from everyone else. And I like this one that he he. Harpreet: [00:30:17] We call on all of us out. Harpreet: [00:30:20] All of us in terms of, you know, people who have, you know, are now if you're watching on LinkedIn, chances are that you are one of these senior Data scientists. And let's look at what they're saying here. Some people apply for 50 plus Data science jobs to eventually land one. Follow up with those people a year or two later, and they're successful in that role. What does that say about the hiring process at all the other companies? The frustration you have and I share it is that those companies don't seem to care. I've been talking about this for over four years now. That means the people who first connected with my post about hiring dysfunction are now in a position to do something about it. So. Just because you went through some bullshit interview round as a Data science candidate doesn't mean you need to perpetuate that nonsense. If you are a senior data scientist now, if you're a hiring manager now and all of us, you know here, if you're up and coming, you will be in that position Harpreet: [00:31:11] One day it is on Harpreet: [00:31:12] You to help break this, this ludicrous cycle. But yeah, are you still angry about entry level job that ask for three years of experience? Harpreet: [00:31:20] Do you remember how ridiculous the Harpreet: [00:31:22] Interview game was? Do you do people deserve to have to wade through the same broken process you did? No, they don't start pointing out the obvious Harpreet: [00:31:30] Dysfunction and Harpreet: [00:31:32] Simple solutions. So then shout out, man, thank you for a thank you for posting that. Harpreet: [00:31:39] But yeah, I mean, look, Harpreet: [00:31:40] Yes, the hiring process is broken. It is difficult, and I think it's just, you know, people perpetuating the same stuff that they went through when they were trying to get their first job and Harpreet: [00:31:52] Just like hazing, Harpreet: [00:31:54] You know, the new sort of it feels like it's just like, feel like I'm getting hazed. Asha [00:32:00] says the best thing I learned from here is search for a job using skills instead of titles. Yes, that's absolutely true. Yeah, you can search LinkedIn job search, just typing actual skills instead of job titles, and you can find jobs that have you read the job description and you're like, Oh my God, this is the data scientist. But then the title might be something different. Uh. I'm keeping track Harpreet: [00:32:26] Of what's going Harpreet: [00:32:27] On on LinkedIn and on YouTube. Shout out to Jeremy Jeremy Ravenel. He's the man behind Nas. Jeremy says, totally in sync with the approach. There's so much stuff that you do with open data. He even gives us a template for open, open weather map. So take a look at that mental gymnastics for data science. That's what's up. Harpreet: [00:32:48] John Withrow, Annette George, my coworker Nanette, how's it going? Harpreet: [00:32:52] I love the transparency. Conversation is useful for many people looking for roles and data science and ml right on. We got Baba in the building. Baba, how's it going? So yeah, Melissa, I'm taking questions. You know, it's not just me ranting this hour, even though he gave me an opportunity to rant. I just keep going. Harpreet: [00:33:11] I dig myself a hole. Harpreet: [00:33:13] If anybody got questions, let me know. I know Matt Blaser was. I needed a little bit ago. I think, Matt, you wanted to chime in Harpreet: [00:33:20] By all means. If you still got that thought, Harpreet: [00:33:22] Please do share it with us. Does not look like it. Asha, go for it. Leah: [00:33:34] Can I chime in on the I want to chime in on the job application process? Harpreet: [00:33:38] Yes, absolutely. Leah: [00:33:40] Definitely apply, even if you're not qualified. Harpreet: [00:33:43] One thing I think of of the interview Leah: [00:33:45] Process like football, right? A lot of the interviews it's practice for when you actually get the job, so even if you don't get the job, the interview will point you to something you didn't know. The more on goal attempts, there will be a goal eventually. So [00:34:00] increase the on goal attempts. In this case, the interviews, and eventually it'll come to just apply the way. Harpreet: [00:34:08] Yeah, I've learned the most from interviewing than I ever did learning on my own because I mean, I don't I don't look at the tedious take home assignments or any of that stuff as challenges I look at as learning opportunities like, OK, like this is something I should probably learn how to do. Like, I've grown more because of the interview process than than any other Harpreet: [00:34:31] Aspect of it, because it Harpreet: [00:34:32] Just forces me to learn new things, right? Harpreet: [00:34:34] Let's say I'm going to work at like I Harpreet: [00:34:36] Remember, I applied for Data engineer role. I don't know why, but this was at the very beginning of my job search and Data science applied for Data engineer role. They gave me a take home assignment, and I was like, Harpreet: [00:34:46] Fuck, dude, I don't know how to do this. This is difficult. I figured it out, and now, like Harpreet: [00:34:52] Now it's like, oh, man, like I could do that task over and over again. Not only that, but I've got something that's now part of my arsenal. I've got code that I've written that I've, you know, referred to and reused, and it's formed a kind of building block for more understanding, right? I've applied for deferred jobs like. Now we're using NLP, I didn't know a damn thing about NLP, but because I went through the interview process because they Harpreet: [00:35:15] Mentioned, Harpreet: [00:35:17] You know, they mentioned certain things and the job Harpreet: [00:35:19] Description, I treated that like it was Harpreet: [00:35:20] A syllabus and I used that as a study guide, right? So if you approach the job search process like Harpreet: [00:35:25] This, every single Harpreet: [00:35:27] Interview you got lined up, look at that particular job description. As ridiculous as it might sound Harpreet: [00:35:35] And look like, just use Harpreet: [00:35:37] It as a syllabus and use that to guide your learning process right? And then give yourself a crash course in that particular thing. Now, if you treat every single one of these things as an opportunity to learn and grow and get better than it's just it's just a game. It's all fun, right? It's not a challenge that you have to go through. It's an opportunity that you have presented in front of you to learn and grow and get better. Like people are telling me, like, these are skills [00:36:00] that we Harpreet: [00:36:00] Like and we want Harpreet: [00:36:01] You to demonstrate them. And now you've got actual like, you've got an actual like learning Harpreet: [00:36:06] Path like, you know what I Harpreet: [00:36:07] Mean? Like, you're just not randomly choosing things. Harpreet: [00:36:09] You're kind of being guided Harpreet: [00:36:10] By what the market wants. Hopefully, that's making sense. Mayor and go for it. Marion: [00:36:19] I just want to mention something from my own experience. I think some Harpreet: [00:36:23] People actually get discouraged in Marion: [00:36:25] The very beginning. There was a period when I was applying for positions and Harpreet: [00:36:30] Was never invited to an interview. Marion: [00:36:34] Uh, so and this put. What extend to three months and more? Harpreet: [00:36:41] So the worst Marion: [00:36:42] People are just starting. That means that we haven't Harpreet: [00:36:46] We don't have a good Marion: [00:36:49] You don't you're not a strong Capricorn. There is something missing in your application Harpreet: [00:36:54] That people just don't pay attention to because Marion: [00:36:58] It doesn't matter. Most likely because that's what to shout to the to the people who post that position, so make sure that Harpreet: [00:37:08] You have a Marion: [00:37:09] Pretty strong presence if you're if you're new to the edit to the Harpreet: [00:37:15] Industry. Marion: [00:37:16] That sounds like I am transitioning from something else. Yeah, my dad's most like, Hey, I'm trying to get into data Harpreet: [00:37:24] Science, but I have 15 years experience and Marion: [00:37:27] Something else, and I have worked in this and this and this. What those companies. So from my experience, when that doesn't me, nobody cares in data science. If I wanted five different wireless communication companies and thought I could've done that, yeah, they can see that they have created all that created Harpreet: [00:37:47] Millions in profit for the companies, Marion: [00:37:50] But that doesn't tell them much. So you need to be more specific. And at the moment, when I changed my life completely to exclude anything [00:38:00] from my previous experience, just mentioning the time I experienced professional 15 years at the very bottom of the list and my designs and whatnot created millions of profit for the companies. But the top is your skills in Data science. So let's say, you know, Python, you know, SQL know everything, you know, Tableau, you know anything, then what we have done in that sense for some projects and describes the project. So those kind of problems to solve. Once I changed my method, Harpreet: [00:38:33] I started getting Marion: [00:38:35] Invited to interviews. So don't get discouraged that you are not getting Harpreet: [00:38:40] Invited to introduce Marion: [00:38:41] Because the point is very Harpreet: [00:38:43] Valid is to Marion: [00:38:44] Get invited to an interview. Harpreet: [00:38:45] You learn from that. You're not going to Marion: [00:38:47] Pass the interview Harpreet: [00:38:48] The first time you're not going to interview. Marion: [00:38:51] The second time, not even the tenth time, because there are always new things happening to you. Couple of days to go. And the test was SQL basically testing my surgical skills. I thought that I'm pretty good at school. It turned out that I didn't know a Harpreet: [00:39:10] Couple of Marion: [00:39:11] Details. I didn't know a couple of the things that they're asking me how to do. So I need to go back and the grocery store, so it takes a couple takes. Working on that and learning new things. But the first thing is you need to get invited Harpreet: [00:39:27] To an interview. If you don't Marion: [00:39:29] Get invited to an interview, you're not going to learn what is expected from you. And the first step is the Harpreet: [00:39:35] Most difficult, but it's Marion: [00:39:37] Very easily fixable. Just do something with projects, put the projects on your resume, and that's a first step. Harpreet: [00:39:46] Thank you very much, Mary. Another question coming in from Stern s on LinkedIn. I am an aspiring machine learning engineer currently taking a boot camp. What are the qualities the hiring manager would look in order [00:40:00] to hire an entry level and male engineer with no experience? Well, you have to think like an engineer. Why do you have to think like an engineer? Because a huge part of your workload is building systems that use Harpreet: [00:40:10] Analytical methods to Harpreet: [00:40:12] Solve business problems. And so to create systems, you need to think like an engineer. So, OK, great. What is the system? There's just a set of connected things that form a Harpreet: [00:40:22] Complex hole, in particular Harpreet: [00:40:24] A set of things working together as parts of a mechanism or an interconnecting network. So you need to demonstrate that you can write software. That's the key thing. And not only that, you need to demonstrate that you can write software that's reproducible because reproducibility Harpreet: [00:40:40] Is a must, right? So make sure that whatever project Harpreet: [00:40:44] You have that you are designing a system, but then you Harpreet: [00:40:46] Also want whatever Harpreet: [00:40:47] Project you have to to be testable, extensible, portable, robust, Harpreet: [00:40:53] Reliable and efficient, Harpreet: [00:40:54] Right? So whatever project you have and make sure you're not writing spaghetti Harpreet: [00:40:57] Code copy pasting Harpreet: [00:40:59] Code from cell to cell all over the place or having your notebook littered with ad hoc solutions or script written with ad hoc solutions, right? Because data science is software right, machine learning engineer is a software position. It might not necessarily well. I'd argue that data science is not software engineering, but machine learning engineers. That is software engineering, so demonstrate that you can think in systems that you can work in systems that you can write. Code that is organized Harpreet: [00:41:30] With documentation is Harpreet: [00:41:32] Reproducible, is reusable. I think that would be one of the qualities that I would look for. So just being able to organize the repository, you have a place where your data lives, place where your documentation lives, place for your visualizations, model outputs, live place your Jupyter notebook lives place for references, a place for source code, place for helper files. Got it all organized in a nice, neat repository. I. So Code Harpreet: [00:41:58] Quality Code Organization [00:42:00] demonstrate he could do Harpreet: [00:42:00] That, demonstrate that you can write code that's clean, has a logical structure, and it's modular. That means, you know, it's easy to beat this system that we're talking about when your code is organized and when your code is clean. It means that anyone that comes to your project to look at it, they could begin to understand what you're Harpreet: [00:42:28] Doing without digging Harpreet: [00:42:30] Into a lot of extensive documentation, right? That is some of the things that. Harpreet: [00:42:39] Look, for biggest thing is just the system's Harpreet: [00:42:41] Mindset, just thinking it systems. He says that makes my software engineering Harp so happy. Talk to us a little bit more about that. What else should a? Machine learning engineers, Spirent, what are some qualities that you would look for that would make yourself to engineer heart happy? Leah: [00:43:05] Well, I think you really hit the nail on the head about just kind of being organized and knowing that you're going to be part of a part of a team and having like a certain level of. Having a tidy code base, having your things organized and putting thought into it because you you'll come back to it later or you'll hand it off to another person and you know, as part of a system, it's one piece in a system. So it needs to be kind of a certain level of professionalism and being polished and finished and also just easily understood. So if someone's like looking around trying to find, well, where's the Data where, you know, even having well thought out like file names or variable names? And that sort of thing is kind of those little things of being doing, doing your future self, a favor about having good organizational skills, about how you write [00:44:00] your software and how you set up your repository. You know that that's so helpful. It helps you and it helps other people on your team. So I think that is a very good thing to kind of keep in mind that, you know, don't be like, ashamed of your code or don't think that, oh gosh, I'm not. I'm not that organized. Just, you know, just work at it and make those improvements. Think about, think about the other people. They'll be working with you as a team. And even if that's not super motivating, think about your future self. If you come back to that project like a month later, help your future self out by having it set up to where you Harpreet: [00:44:32] Can quickly understand what you Leah: [00:44:33] Were doing. Harpreet: [00:44:35] Yeah, 100 percent. And that's the nice thing about Python is that it's like Harpreet: [00:44:40] So readable like by humans like, you know Harpreet: [00:44:44] That you can go back and kind of understand what you're doing, even without good documentation, although you should always have good documentation for your code. I think that, yeah, it was Travis Oliphant, creator of Nampai and Sci-Fi. I was listening to an interview with him on the Lex Friedman podcast, and he's talking about what drew him to pot. Sorry, what drew him to Python was the fact that he revisited what he did a couple of years later and was like, Oh, I still understand this stuff. I mean, that's huge. That's key. So hopefully that was helpful. Harpreet: [00:45:16] Stern. Let us know if you got Harpreet: [00:45:17] Any follow up Harpreet: [00:45:18] Questions. Harpreet: [00:45:19] Jeremy on LinkedIn. Triple thumbs up on the systems. There you go, man. Triple thumbs up on the systems. All right, so if anybody gets questions, let me know. Shout out to Bob and Chin, may you guys right here who are Harpreet: [00:45:38] In the room. Cameras are off. That's OK. If you're Harpreet: [00:45:42] Shy or shy, Harpreet: [00:45:43] It's all good. But let me know if you guys got questions. Harpreet: [00:45:46] I'm keeping an eye out on YouTube and on LinkedIn. If you're watching this on Harpreet: [00:45:50] Linkedin, by the way. Harpreet: [00:45:51] Smash a reaction. Make sure you share this with your network. Help spread the word, man. We're doing these Sunday office hours for a while [00:46:00] on Sunday until we start moving them to a different different. Date and time. Great advice here from Harpreet: [00:46:09] Barrett on the Harpreet: [00:46:11] Linear note, yeah, definitely. The full stack deep learning course is amazing. Harpreet: [00:46:17] I've not taken it Harpreet: [00:46:18] In its entirety, but. Definitely is a Harpreet: [00:46:21] Good course, I think it is completely Harpreet: [00:46:23] Free if I am Harpreet: [00:46:25] Obviously pretty sure it's free. I know Makiko, a Harpreet: [00:46:29] Good friend of mine, Kaus is big on on full stack, deep learning as well. And while you guys are out there and make sure you're managing your Harpreet: [00:46:40] Experiments, right? Harpreet: [00:46:42] Experiment management is key. A great way to manage your experiments using comet. I might add I. You can comment is a good way to manage experiments real, though, it really is, and it's completely free for for free guys to sign up as an individual. So if you go to Comet Harpreet: [00:47:02] Ml, let me pull this up a quick. Harpreet: [00:47:05] You guys can go to comment on ML, and here's what it looks like when you log in, right? Like I was just playing around a little bit. Here's a project and this is what what email looks like. It shows how my batch loss is comparing against each step, and I can look at different like different metrics as well. And I'm tracking all of my experiments, tracking the different parameters and what the output was and Harpreet: [00:47:31] Logging artifacts and stuff like Harpreet: [00:47:33] That. I can go here, look at a confusion matrix I could see. You know what was wrong? Ok, well, here's a zero that was misclassified as an eight, right? Here was some list classifications. This was misclassified as the three. Harpreet: [00:47:45] So it gets Harpreet: [00:47:46] Really, really. Like really good feedback on uncomment. You can also track metrics as well. Another great thing about it is. All your code and then the artifacts [00:48:00] as well. The facts being the Data, so. Like the Data in this one, but you can log your Harpreet: [00:48:09] Data as well. You can comment so version controls Harpreet: [00:48:12] Your Data as well as your code and your parameters and your models. Definitely check out comment. Ml If you have not already, you'll be hearing me give more presentations on that. We'll be we'll be pumping out more content with more interesting use cases if you are listening. If you're here, if you want to see some interesting stuff, get done. Any ideas Harpreet: [00:48:34] For a project you'd Harpreet: [00:48:35] Like to see worked out? Go ahead and let us know. You can either send me an email at Harpreet Sahota Comet Ml or comment right here on this video, and we'll take note of that man. Let me know if there's stuff that you guys want to see. Get done. Be happy to do it, man. Any type of projects. Any of the questions coming in? Let me know. Let me know our friends. Jeremy on LinkedIn says exactly Python is so clean to read, I went to Python after years of Excel and VBA in finance. Yeah, I was using a lot of VBA back when I was an actuary and it was not. Harpreet: [00:49:15] Wasn't that intuitive, I would say. Harpreet: [00:49:18] It's very different, VEBA remember trying to write the Black Shoals formula Harpreet: [00:49:22] In VBA to, I think, some options pricing, Harpreet: [00:49:25] And that was quite challenging. That run in a spreadsheet. All right, we got we got some new people in the room, actually. Colorado, what's going on shouldn't be part of the bubble. Look, man, if you guys got questions, now's the time to ask you a question, man. Harpreet: [00:49:41] I know you guys didn't Harpreet: [00:49:41] Just come in here just to hear me rant. You got questions. I might not have an answer, but I could probably point you to the right resource. So go for a man, Harpreet: [00:49:50] Akshay or Baba or parrot. Harpreet: [00:49:53] I know Tim is not connected to audio, so that's fine. Harpreet: [00:49:57] But if anybody ask questions, let me know. I'm also looking Harpreet: [00:49:59] On [00:50:00] Harpreet: [00:50:00] Linkedin and YouTube, Harpreet: [00:50:01] Otherwise I can start to wind it down. 16 people watch it on LinkedIn now 15. Now 14, it is dropping my friends. Can I have the officers, that's your questions. I'll pause. [00:50:13] I'm speaking of air. Harpreet: [00:50:19] All right, all right. So, Leah, talk to us real quick about your Harpreet: [00:50:25] Data analytics Harpreet: [00:50:26] Data Harpreet: [00:50:26] Scientist by way of software engineering, what was your trajectory Harpreet: [00:50:29] Like to transition into data science? Leah: [00:50:36] Well, for me, it was like finding a random course on Coursera, like years ago, I was doing software engineering and I found this cool class on like doing that analysis from Johns Hopkins on Coursera. And I started to do it kind of for fun, and I also started to do more things with using statistics. Harpreet: [00:50:57] And even though Leah: [00:50:58] I've been doing software, I was kind of a math phobic and statistics was like Harpreet: [00:51:04] More a nicer Leah: [00:51:06] Math. You know, I was like, Oh, I Harpreet: [00:51:07] Could, I could. I could. I don't feel Leah: [00:51:08] So intimidated about statistics, so it just interested me. And I kind of did that as a hobbyist. And recently I had the opportunity to do a Data science bootcamp. And I thought, OK, you know, I'm just going to do a career. Not not a huge shift, but a little pivot from software into into Data. So, so now I'm looking more at taking my software skills and using that to build this foundation for doing more with data analysis and with with the data science. And I do get hit up for Data engineer jobs as well. So I was like, Well, that makes sense because it seems like a natural kind of progression. So that's what made the made the shift. Harpreet: [00:51:56] Thank you so much for sharing your experience, Leah. Harpreet: [00:51:58] Yeah. I mean, Harpreet: [00:51:59] Stats [00:52:00] is awesome. Like, I mean, I love probability theory Harpreet: [00:52:04] More than Harpreet: [00:52:05] The rest of the Harpreet: [00:52:05] Statistics, but statistics Harpreet: [00:52:07] Is useful and it has its place. All right, Baba is asking two questions, all right, Harpreet: [00:52:12] For a fresher how to prepare for Harpreet: [00:52:14] Computer vision Harpreet: [00:52:14] Roles. All right. Harpreet: [00:52:18] I mean, first I would I Harpreet: [00:52:19] Would I would here here's a great. Harpreet: [00:52:21] Resource, I'm going to link you to real quick. Check this out or to understand what computer vision is and what it is used for. So if you don't know what computer vision is, how it's used, then start here six significant computer vision problems solved by ML. You can look through here, and the first thing is is, I mean, you need to get your hand on some good learning Harpreet: [00:52:40] Resources, right? Harpreet: [00:52:41] So a couple of good places I would look to as I think, my friend John Krohn. Has some resources on this. I thought he did, but he did not. So. I mean, just to get started is just to understand what a CNN is and how does the CNN work, those convolutional neural networks, I think that is probably foundational for poor computer vision. So just get an understanding of that. So I would recommend a couple of really good books here if you're interested in getting. I mean, if you're a fresher and you're jumping right to computer vision without having a foundation in, you know, other data science topics, probably not going to set yourself up for success long term. That being said, that warning being given, this is a really good book right here. Harpreet: [00:53:33] Deep learning illustrated, right? Harpreet: [00:53:36] This has an entire section here on computer vision, and it's written completely without any. There's no math. Everything is Harpreet: [00:53:47] Kind of taught visually Harpreet: [00:53:48] And intuitively. So it is a very good read for that. And they have. Like three chapters on convolutional Harpreet: [00:53:58] Neural Harpreet: [00:53:59] Networks, I think [00:54:00] definitely check that out. Another good book is a deep learning illustrated, so it's a good book as well. Definitely check these out. Yeah, I mean, I didn't answer your question because like, I don't really have like a Harpreet: [00:54:13] Full path for you. Harpreet: [00:54:15] I think that would take a lot more back and forth conversation. Harpreet: [00:54:19] But do you keep Harpreet: [00:54:19] In mind that in the near future at Comet, I will be having like a learning journey for computer vision, so, you know, keep an eye out for that. And hopefully my useless answer Harpreet: [00:54:31] Was not that bad. Harpreet: [00:54:33] How important is skill? How, how to master it? Ok, well, sequel Harpreet: [00:54:37] Is important, right? Harpreet: [00:54:38] So now you're asking two unrelated questions Are you trying to do computer vision or are you trying to do a sequel? Because I'm sure that people who are working in computer vision probably don't need to use much skill. Harpreet: [00:54:46] Right? So, so there's a Harpreet: [00:54:49] Disconnect there, right? So I don't know many computer vision experts who are using sequel on a day to day. Maybe they Harpreet: [00:54:55] Are, because computer Harpreet: [00:54:57] Vision inherently deals with unstructured data where a sequel is very Harpreet: [00:55:01] Structured Data. So either way, sequel is important. Harpreet: [00:55:06] You need to know how to do it, but it's not hard to Harpreet: [00:55:08] To get good at. Harpreet: [00:55:10] How do you master it? Start learning it. I mean, how do you master anything you practice? You just do it over and Harpreet: [00:55:15] Over and over, right? Harpreet: [00:55:17] You don't just listen to lectures and that and just have that, be it right? Harpreet: [00:55:22] You learn skill by just doing it straight up, doing it. Harpreet: [00:55:27] Um, Data scientists currently in India. My advice is completely useless in India. Harpreet: [00:55:32] You know, I might look Indian as I am, Harpreet: [00:55:35] But I've never had to go to school in India and never had to work for in India. Never had to look for a job in India. I'm North American, North America, U.S. and Canada is my area of expertize and most of the people I've helped are in North America. Harpreet: [00:55:48] India is an Harpreet: [00:55:49] Entirely Harpreet: [00:55:49] Different country culture Harpreet: [00:55:51] Value system and you have literally 300 times the population of Canada, so Harpreet: [00:55:58] You all got more Harpreet: [00:55:59] Competition [00:56:00] just looking for the same jobs that I would be looking for. So Akshay says not sure Harpreet: [00:56:08] What should be the right path. Harpreet: [00:56:09] Studying abroad is one option, but how good our chances of job hunting at a global level since a lot of companies can't offer sponsorship. You don't know the right path can be for you, either, my friend. Harpreet: [00:56:22] That's a huge Harpreet: [00:56:22] Decision that you know you shouldn't trust Harpreet: [00:56:26] Strangers Harpreet: [00:56:26] Like really think about it, right? Like, what is it that you want to accomplish, right? Like. Like or what is that you want to accomplish in life, is it necessary for you to leave India and pursue higher education somewhere else Harpreet: [00:56:42] In the world, right? Harpreet: [00:56:44] I don't know what it is you're trying to do with your life. I don't know what you're trying to accomplish, but Harpreet: [00:56:47] Just to ask yourself that question. Figure out what it is you want to do. What does he Harpreet: [00:56:51] Want to accomplish and then ask yourself for me to do this thing or accomplish that thing? Is it necessary for me to go to school Harpreet: [00:56:59] Or live or Harpreet: [00:57:00] Work in a foreign country? Right? If the answer to that is no, then get your ass Harpreet: [00:57:05] To work where you are and start making that thing happen. Harpreet: [00:57:09] That being said, actually, I see your ID. Speaker5: [00:57:13] Yeah, hi Harpreet. So, yeah, so I was not speaking earlier because there was a noise in my background. So it's always it's not always clear. Absolutely. Yeah. So my question Harpreet: [00:57:24] Here is that Speaker5: [00:57:26] So I'm going to be working in India as a data scientist. Ok, so I want to export opportunities abroad, as a data scientist or as a machine learning engineer. And I see that there is a lot of difficulty when you go and find a job at a global level because Harpreet: [00:57:39] The companies can't Speaker5: [00:57:40] Afford visa sponsorship. And yeah, I actually do not want to do a graduation from any of. And if any, foreign university, so I just want to explore opportunities at a global level. So my question was that how should I say it or what should be my path Harpreet: [00:57:57] For this job hunt? Harpreet: [00:58:00] I [00:58:00] don't know, man, I can't give you advice on that because that's not a challenge that I've had to face my life ever, so I'm not going to be the best person to ask for that. Harpreet: [00:58:09] So I really Harpreet: [00:58:10] Can't help you in that respect, man. Like, I'm going to say stuff that's Harpreet: [00:58:12] Going to be inapplicable, Harpreet: [00:58:14] But maybe Marion can. Marion: [00:58:17] I don't know how relevant it is, but Harpreet: [00:58:19] When had a couple of years Marion: [00:58:21] Ago looking for what we need to do to learn that Harpreet: [00:58:25] And learning? And the first Marion: [00:58:27] One of that Harpreet: [00:58:28] Was that that was Marion: [00:58:30] Actually how successful candidates in getting a job in different countries say, I forgot the Harpreet: [00:58:39] Link. It was a couple of years ago, Marion: [00:58:41] But that was a very good comparison between people with no experience or very little experience applying for that job. And that's what struck me was Harpreet: [00:58:52] In the US, it was Marion: [00:58:54] One of the Harpreet: [00:58:56] Maybe below 10. Marion: [00:58:59] One out of 10 or one out of 20 will get the job in the UK, it was three out of 10. So do some sort of research while the market. The demand for data scientists is really good and there is not that much supply. Harpreet: [00:59:16] From my Marion: [00:59:17] Understanding. The United States is actually probably the hardest country to get into Data science. Your story is different because there are all scientists, but still the competition is very, very strong and it's difficult to break through. I'll think somewhere, somewhere in Europe, it's probably easier. Yeah, and that's what I remember my my head was like, wow, the United States, Harpreet: [00:59:47] You have less than Marion: [00:59:49] One tenth of one in 10 chance to get the job. If you're just starting Harpreet: [00:59:53] In the UK, Marion: [00:59:55] Three times better than that. So if I had to choose [01:00:00] and if I had the option to go to the UK, I would do it in a heartbeat. Harpreet: [01:00:05] So this is my Harpreet: [01:00:06] My my Harpreet: [01:00:07] Stance on Harpreet: [01:00:08] This right? Harpreet: [01:00:08] And you take it with a grain of salt like whatever. Harpreet: [01:00:11] I mean, Harpreet: [01:00:12] If there are qualified candidates for a role who are citizens of your country, then you should probably hire those people first, right? I think that companies that are based in a certain country have a duty and an obligation to hire their own Harpreet: [01:00:26] Citizens first, right? Harpreet: [01:00:27] And then maybe after their Harpreet: [01:00:28] Citizens permanent residences, right? Harpreet: [01:00:32] That's just just my philosophical standpoint. Now that being said. Yeah, it's difficult for anybody to come into a new country, try to find a job, right? If that's a difficulty you want to place yourself in, then definitely go through that path and put yourself in that difficult situation. Or can you ask yourself, are I? Can I make a big difference in my own country with my own skill set, right? And it seems like India is a place that could Harpreet: [01:00:54] Do that is technologically Harpreet: [01:00:56] Advanced in some places like I don't I don't know much about India. I don't know why you want to come to USA. Harpreet: [01:01:00] Maybe you just think it's cooler Harpreet: [01:01:02] Here, but can you do something big in your own country with the skills you have and make Harpreet: [01:01:06] An impact there, right? Harpreet: [01:01:10] So, yeah, I don't I don't I don't know anything about having to job hunt on a global level. Look, man, I'm North American. My woes have only ever been in North America, so I'm just speaking from that point of view. Parathas asking, Do we really need a masters to break into Data science abroad? Got, I don't know. I don't know if you really need a master's to break into data science abroad. Software engineer in India, I want to break into Data. I want to break in to data science jobs in the U.S. So let me ask you this part like why not just try to break into a data science job in India? Harpreet: [01:01:41] So like what? Harpreet: [01:01:43] I mean that this question might you know? Let's do see why why the U.S., why do you want to move to the U.S., like why do you want to leave India and become a Data sign? If you're already Harpreet: [01:01:53] A data scientist in India, Harpreet: [01:01:55] Why do you want to become a data scientist Harpreet: [01:01:56] In the U.S.? I guess that that would Harpreet: [01:01:58] Be my question. I'd [01:02:00] love to hear from party orthodoxy on that. Speaker5: [01:02:05] Yeah. So. You go ahead, but at first. Harpreet: [01:02:10] Right, exactly. So my quip Harpreet: [01:02:13] Is first, Harpreet: [01:02:14] They want to break into Data Harpreet: [01:02:16] St and like Harpreet: [01:02:18] The U.S., is just an added bonus, and Harpreet: [01:02:21] It's not like Harpreet: [01:02:22] A priority to break to get into the U.S.. I'll make that clear, though. I bet I'm having trouble breaking Harpreet: [01:02:30] Into it in India itself, and I might Harpreet: [01:02:33] Be having some misgivings or misconceptions Harpreet: [01:02:35] That the competition in the US is really lesser, but Harpreet: [01:02:40] There's the competition. The U.S. is going to be far more Harpreet: [01:02:42] Difficult because now you're Harpreet: [01:02:44] Going against people who are graduating from the absolute best universities in the world because most of the absolute best universities in the world are in the U.S. So if you think it's going to be easier to get a job in the U.S., I think you as a foreign, it's not. It's going to be more difficult. Absolutely more difficult. That's for sure. So anyway, at this, probably something that we could talk about on my own personal office hours, not in this one. So let's let's go ahead and go ahead and move past this question. Just I just want to put that seed for thought there. Like why? Why not just, Harpreet: [01:03:22] You know, try to Harpreet: [01:03:22] Make things better for yourself in your country than than try to move to USA? I don't know. Like I might get my get a lot of heat for even posing that question, but I think it's something that you guys should definitely think about. If you are under the misconception that will be quote-unquote easier for you to get a job in the USA. I think you are absolutely wrong. It'll probably be more difficult because first of all, you have the visa issue to face, and that in itself will make it in relatively difficult challenge to undertake. And our last question we're going to take from LinkedIn is from Derek. Harpreet: [01:03:56] What would you Harpreet: [01:03:56] Say the best practice is for mentorship? Do you recommend having [01:04:00] mentors from multiple industries and multiple disciplines or mentors in your industry and discipline? So best practice for mentorship is just view anyone you encounter as a mentor. You don't need to like, actually like, sit and talk with that person for that person to be your mentor. Some of my best Harpreet: [01:04:14] Mentors, for example, Harpreet: [01:04:16] Naval Ravikant, I consider him to be a mentor Harpreet: [01:04:19] Of mine. Harpreet: [01:04:19] Never spoken with the guy. Probably never will. But I take a lot of knowledge and from him and reflect on that. So the best practice for mentorship is just don't assume that it needs to be a one on one game. Take the opportunity to interact with Harpreet: [01:04:34] With ideas Harpreet: [01:04:37] From different times. Even I read old books look to people who are currently alive, that you have written books and read their books and get their ideas right. Because if you read a book that somebody wrote, you're downloading decades in days, right? Highly recommend having mentors from multiple industries, absolutely right. Harpreet: [01:04:58] So like for me, a lot of my quote Harpreet: [01:05:00] Unquote mentors are angel investors, philosophers, physicists like the only mentor I have in Data science. Harpreet: [01:05:09] I mean, not the quote unquote only, but I've got many Harpreet: [01:05:11] Mentors in Data science, but my biggest mentor and Data sciences is probably just like, Harpreet: [01:05:17] Hands down. My biggest mentor would be even, Harpreet: [01:05:20] You know, also followed by Tom Harpreet: [01:05:21] And Joe and and Ben. Harpreet: [01:05:23] And all Harpreet: [01:05:24] Those guys are those, those guys are all my Harpreet: [01:05:25] Friends. Harpreet: [01:05:26] I love them a lot. But yeah, I'd say Harpreet: [01:05:28] Vin would be like the person I would consider my Harpreet: [01:05:30] Mentor, but I don't Harpreet: [01:05:32] Have mentoring sessions with him, like I don't like, regularly meet up with a van and talk to him about my issues and my problems. I just look at what he's Harpreet: [01:05:40] Saying in various Harpreet: [01:05:41] Interviews. I read his post and I reflect on it, and Harpreet: [01:05:44] Then I kind of merge that Harpreet: [01:05:47] In with my own thinking and try to find ways to connect ideas to my own experience from what he's saying. I think that probably the best way to Harpreet: [01:05:54] Use a mentor. Harpreet: [01:05:58] It does not look like there are any [01:06:00] other questions, my friends, thank you so much for Harpreet: [01:06:02] Joining in here. Harpreet: [01:06:03] Really appreciate you guys taking time to come and hang out. Great questions, great sessions. And as usual. Remember to tune in to my podcast, the artists of Data Science. Very proud of that podcast. I think it is phenomenal and I think you guys will enjoy it as well. Harpreet: [01:06:19] So tune Harpreet: [01:06:20] In. Let me know what you think. Harpreet: [01:06:21] And as usual, Harpreet: [01:06:23] My friends remember you've got one life on this planet. Why not try do some big cheers, everyone?