Andrew Jones_mixdown.mp3

Andrew: [00:00:00] You know, putting the nuance on that to make it actually work. I think for me as well, like auto and mail doesn't have any business sense necessarily, so it doesn't know what problems to solve or it doesn't know why it should solve them. So I think humans are still a huge part of that. I don't think that's going away anywhere soon. It's just an evolution and data scientists are going to start, you know, there's going to be bits where automation comes in and helps us do our jobs even better. But I don't think it's going to take away jobs necessarily.

Harpreet: [00:00:38] What's up, everybody? Welcome to the Artists of Data Science Podcast, the only self development podcast for data scientists you're going to learn from and be inspired by the people, ideas and conversations that will encourage creativity and innovation in yourself so that you can do the same for others. I also host Open Office Hours. You can register to attend by going to Bitly.com/adsoh forward slash a. Ds0h. I look forward to seeing you all there. Let's ride this beat out into another awesome episode and don't forget to subscribe to the show and leave a five star review.

Harpreet: [00:01:20] Our guest today has spent over a decade in analytics and data science at some of the world's most innovative companies, including Amazon and Sony PlayStation. Over the course of his career, he's interviewed and screened hundreds of data science candidates, and through this process has learned exactly what differentiates and employable data scientist from the rest. He's taken that experience and authored a book titled The Essential AI and Data Science Handbook for Recruitment. And throughout his career, [00:02:00] he's also been a mentor to fellow data scientists from their entry into the field to developing their technical and non-technical skill sets, as well as providing guidance around preparing for and being successful with promotions and interviews. The culmination of that experience has resulted in the creation of data science infinity a new approach to learning data science. His program has taken input from hundreds of data science leaders and recruiters within the field and is aimed at providing you with everything you need from the fundamentals through to landing a great role in this exciting industry. So please help me to welcoming our guest today, a man who is committed to teaching us a better way to learn and do data science. Andrew Jones Andrew, thank you so much for taking time out of your schedule to be on the show today, and I appreciate having you here.

Andrew: [00:02:55] My pleasure. Completely. Humphrey. Woman Intro.

Harpreet: [00:02:58] Wow. Absolutely. Absolutely. Yeah. It's just great to actually get you on the line and actually talk to you. I know we've been in contact. We run in the same circle, so obviously we it's weird, but we're kind of in contact almost on a daily basis through this little group chat we got. But I've never actually like sat down and talked with you, so this is an honor to have that happen. Let's learn a little bit more about you. Talk to us about where you grew up and what it was like there.

Andrew: [00:03:25] Well, yeah, I mean, I so I'm in London now. We're just out of London now, but have been in London for a lot of the last ten or 12 years. But I, I originally come from small town dairy farming, New Zealand, you know, like a few thousand people in the town, you know, one primary school, one high school, pretty different from London. And I, I grew up there from, from age zero, right up to going to university. I studied in Wellington in New Zealand and then worked for a couple of years and then decided that I wanted to see what else was out there in the world and train traveled the world and ended [00:04:00] up in the UK. I actually have a UK passport which makes it a lot easier. My my parents are originally from the UK, so that made that transition super easy. But you know, in terms of coming from super small town New Zealand, I think, you know, the place they came from was an awesome place to grow up, you know, a pretty simple existence. But in in hindsight, I think now I'm very, very glad that I've gone and seen the world and lived in a city like London where, you know, every person you talk to or work with is from a different country or has a different accent or a different skin color or faith or different beliefs in general.

Andrew: [00:04:39] It sort of forces you to become, you know, there's only one way, and that's to be a very open minded person. And it means that you avoid falling into the trap of being huddled close with your own type and being fearful of things that are different. Because, like I say in London, like any sort of metropolitan city in the world, everything and everyone is different. There's no avoiding that. And I absolutely love that. And I think that's a very important thing, especially with everything that's going on in the world at the moment. There's a lot of division. You know, I think that idea that people can be different from you and just dealing with that day in, day out is such a good thing to to be immersed.

Harpreet: [00:05:16] And yeah, I've actually I've been to New Zealand once. My wife and I were we did a our honeymoon across Australia and New Zealand within three weeks, went from a Sydney, did a road trip on the coast to Melbourne and then from Melbourne flew into New Zealand, into Auckland and absolutely loved Auckland man. Like it was a cool place but like I felt like it was a bit multicultural there as well up in Auckland, but Wellington's at the South Island.

Andrew: [00:05:43] Wellington's at the very bottom of the North Island. So if you were going to the South Island you would leave from Wellington on the ferry to go across to the South Island. Wellington's like the third biggest city in New Zealand, but it's quite a lot smaller than Aucklanders. Probably got maybe 400,000 [00:06:00] people and I guess whereas Auckland's Auckland's a bit bigger, Auckland's maybe like one and one half million.

Harpreet: [00:06:06] It's a cool place, man. I absolutely love, absolutely love New Zealand. So I mean, London is cool too. Don't get me wrong, I like that place. But man like New Zealand just felt so laid back and chill like it didn't feel as crazy hustle bustle as as London did. So when you're in high school, what did you think your future would look like?

Andrew: [00:06:24] Oh, man, that's a good question. I don't even think I could remember what I thought my future would be like in high school. So, I mean, in terms of what I what I enjoyed studying from a school point of view, like I enjoyed things like math and and economics. So I guess somewhat related to data science in a way. But in high school, like I didn't know about coding or I didn't know about data or anything or anything like that. Absolutely not. I was big into sport. That was probably like my my biggest passion. I probably thought I'd have a career and I probably misguidedly thought I'd have a career in sport somewhere representing the country and cricket or something like that. But I think I think I've passed my chance now of representing New Zealand in sport, unfortunately.

Harpreet: [00:07:12] At six foot seven. It's it's a shame that you did not get into basketball. Is that right? Six, seven. Your height.

Andrew: [00:07:18] Six foot six. Six foot six. Thanks. Yeah, I think I come in at about 199 centimeters. I just missed out on the two meter mark, but I play. Yeah, I played a bit of basketball. I do like I do like basketball, but yeah, that's a good question. I honestly thinking back to what I thought about where I would go, I honestly, I don't think I even knew that something like analytics and data science existed at that point in my life.

Harpreet: [00:07:44] So that's about how you got into into data science and analytics. Like what was that dream like going from? Did you start doing analytics in New Zealand to start in London? Like I guess walk us through that journey.

Andrew: [00:07:53] So my, my sort of journey into, to where I am now is a little bit of an odd one, at least compared to most [00:08:00] other people in the industry. So I, like I say, I studied in Wellington and I did what's called a conjoint degree where you do you essentially do two degrees bundled together in one. So I did a commerce degree majoring in marketing and I did a psych science degree majoring in psychology. So neither of them are particularly stem based subjects, even psychology. There's a little bit of statistics and whatnot in it. But I didn't know anything about marketing. Sorry, I didn't know anything about analytics or data science until I stumbled into the role. So I actually was playing indoor cricket with a group of guys from that I knew through a shared friend, and I was I was graduating from university with marketing and psychology, but, but I loved mathematics. And even though I wasn't studying it and I was just talking to a guy in my cricket team who was a little bit older than me, and he was the manager of a marketing analytics team at a a telecommunications company in New Zealand. And I was saying, look, I'm graduating soon and just off the cuff talking and he said, you should come in and interview for a marketing analyst role.

Andrew: [00:09:03] It sounds like something might be a decent fit. And I went in and interviewed and very fortunately got in from in that first role I learned to code and SAS and I just was blown away with the idea that you could manipulate data and analyze data and understand what customers were doing and and then, you know, delivered above that saying things like predictive modeling. I remember seeing logistic regression model the first time being used to predict which customers were going to churn the next month from the telecommunications company and then the following month, seeing that the customers, which they said were the 10% most likely to churn this huge proportion of them then left the business. I thought it was magic and all this stuff. It just clicked with me, not in terms of me being able to do it, but it just clicked with me in terms of it sort of lit a [00:10:00] fire inside of me. It just it felt like I wanted to do more and more and more of it. So that's that's been my inspiration. I just I love the stuff that I do. And now I get to I get to create content and teach it. This is awesome. But yeah, I absolutely stumbled into it, literally stumbled into it through an indoor cricket team connection.

Harpreet: [00:10:20] I used to code an SAS as well way back that way back in the days wasn't that long ago, but for a rule that I had that was like four and a half years, whatever in that rule. All that stuff was done in SAS as well. Yeah. Yeah. I've got a love hate relationship with that, I guess. I mean, I like Python better, I guess. But what's the toughest part about transitioning from SAS into like a quote unquote? Yeah, let's just say proper programing language like Python.

Andrew: [00:10:48] So I, my sort of transition from SAS was into ah, so I learned ah over here in London and that was a write I sort of, I can't remember how I sort of got into it. I, I started using it for a few projects which I needed at work. But then I discovered Kaggle quite early on and I just loved that idea of trying to build predictive models and competing against each other. But it's obviously cable's grown into a beast now. This was this was before it was massive. And that's how I sort of learned a lot of my coding. And then I actually when I was at Amazon, I was still using mostly, but a little bit of Python as well. And then when I moved to Sony PlayStation, I was doing a lot of prototyping of machine learning based features and the engineering teams there just didn't even they didn't even know what I was. They just they only dealt with Python. So I was I just I just translated the code over to from our to Python to do the things I wanted to do. And that was actually a really good way of me learning Python because it was sort of forced to, you know, I knew exactly the steps I needed to go through to get from whatever the task was I needed to solve to the end product. I knew everything I needed to do. I just needed to [00:12:00] figure out how to do that in Python. And it's it's probably one of the better ways to learn, actually. You don't have to learn from scratch. You're learning from an, you know, a template of knowing what you need to do. You've just got to figure out the syntax you need. But yeah, yeah. So I guess that was a reasonably easy transition, thinking about it that way rather than trying to learn it from the ground up. I think I was quite lucky. Yeah.

Harpreet: [00:12:23] Yeah. That's an important point because when I was going from learning fast to Python, what I did was I would just recreate the work I did in SAS. But with Python code I knew what my SAS output was supposed to be, so I just recreate it with Python and it was like automatically checking whether what I had coded was the anticipated result. So you mentioned going from SAS to, to ah to Python. Is the learning curve for learning a new language become shorter and shorter as you just become more and more better at coding regardless of what the language is.

Andrew: [00:12:55] Yeah, I think I think that's probably fair to say. I think to a certain degree, I think coding in general up to a point. So there's people who are extremely good at coding in whatever language that they use. And I'm not talking about them. I think for the vast majority of coding, that's not the difficult bit, especially with resources like StackOverflow, where you can literally find out how to do anything in a way where the the the solutions are ranked by other people's approval of them. You know, you can find pretty good solutions on something. I don't think that the syntax of coding itself not too difficult. It is more the understanding of what you need to do, the steps you need to take, whether that's like a a machine learning pipeline or just building a machine learning model and understanding what considerations you need to make based on the scenario you're in or the type of algorithm that you were using and the ways that it deals with data differently. That's the hardest bit to learn because you [00:14:00] sort of you only really learn those by trial and error unless, you know, you go through a course or a book or something.

Andrew: [00:14:07] But the coding itself is a lot of sort of template stuff. And what I found over my career is something that I luckily started doing quite early on was I would always save code snippets right back to SAS. I would have a notepad file of everything I'd done, and then everything would be at the touch of a shortcut key to do whatever I needed to do. So I always had this quick template to put stuff into place, or if I'd done something in the past at another business, I'd sort of know the template or the stencil of what I needed to do. And then I just rejigged that to to work for the scenario that I was in. So you never have to sort of start from scratch. You've always got this base of content to work for and that's something in data science, infinity on, big on as well as creating templates for people to then have for whatever scenario they encounter. They've got 80% of what they need there and they just need to move the the sort of edge cases around. And then they've got pretty much their final solution.

Harpreet: [00:15:08] Thank you. I appreciate that. Yeah. I guess for me personally, like, I started learning our way back like undergrad and then from AH went to like BBA and Excel then to fast and to Python. It just felt like. Once I kind of knew how to do something in one language, it just became easier to learn something new just because it's just more of a way to kind of think than it is. Like you mentioned, the actual syntax and stuff like that.

Andrew: [00:15:34] And the obvious is the Iron Python transition was. I mean, like at the very, very granular level, there are some differences, but for the most part, they kind of work in the same way when you're trying to say you trying to build a machine learning problem like the syntax is different. The way they they you input your data and stuff can be slightly different, but it's more it's more like going [00:16:00] from an Android phone to an iPhone. You know, it's like an annoying, subtle difference rather than a major difference of thinking. It's not it's not that different. So, you know, anybody that knows AR but thinks they want to learn Python or vice versa, I think, you know, you could do it. It's just one of those things where you get into such good habits with one language that it becomes annoying, more difficult to move to another language.

Harpreet: [00:16:25] So you've been in this field for over a decade. How far has it come since you first broke into it?

Andrew: [00:16:31] Well, I think I mean, from my point of view, I would say that there have been there have definitely been changes in the industry since I came into it. Obviously, I've progressed into different roles and different industries. So so comparing exactly what I was doing at the start with what I'm doing now is a little bit tricky, but. In general, I'd say there definitely have been changes, but there are things that have stayed pretty constant as well. So like obviously there have been radical increases in the amount of data that's being generated and the number of companies that are generating it. I saw a statistic the other day which it always blows my mind I think I'd seen once before, but it was that every minute of every day 500 hours of YouTube content is uploaded to YouTube. So like every minute of every day, there's 500 hours of video content going up. I mean, that's YouTube, one of the biggest companies in the world know obviously being part of Google, but in video data being very rich but. It just kind of shows you the insanity of the amount of data that can now be stored and used, and that for a lot of companies that's not the case, but that they are starting to collect a lot more data. Companies are looking to capture everything they can because they know that there are ways to now use that to to get ahead of the competition or stay up with the competition.

Andrew: [00:17:51] There's been rapid hardware increases, obviously, when when I was first in analytics with data science, the idea of this [00:18:00] was probably before even deep learning was really possible because GPUs over the past five years have really, really come along and that has come with these massive advances and deep learning as well, especially in light language understanding and language generation. But in saying all of that, there have been these immense increases in data and in hardware. And this is a lot of new, amazing things have happened at the cutting edge. But. None of those things. I would say for the vast majority of businesses, none of those increases and things have instantly solved the vast majority of sort of data related problems that they have. There's definitely been sort of a lot of hype around data science and whatnot for the last few years. But I think people are starting to realize at the end of the day, sometimes, like a simple common sense solution to the actual business problem or customer problem can actually make a lot more sense than the past three years. There's been this this huge emphasis that data science is going to solve every problem that a business has. And I think people are starting to understand what data science is and what data science is. And putting in place a simple but effective approach can be the best way forward sometimes.

Harpreet: [00:19:13] So can you share a hot take with us on where do you think the field is headed? Remember, this is this is the hardest data science podcast. So we all have a little bit of controversy. So maybe sharing a secret contrarian viewpoint that you hold, which might be different from the rest of the data community.

Andrew: [00:19:30] Well, in terms of where data for data science or the data field is going, I could probably say with 100% certainty that I have I have no idea. And I don't think anybody can tell you where it's going, because we're just seeing crazy stuff happening at the cutting edge. We're probably going to see massive further advancements in the areas of language, AI, computer vision, maybe there's a lot of headroom there. We're sort of lagging behind the language models now. I guess my slightly contrarian view on data science [00:20:00] is like I personally have zero fear that AutoML is going to come and take away data science roles. So I think AutoML has its place. I don't believe the hype that it's going to kill the data science industry. So AutoML on its own that it can allow you to, to scale and build huge volumes of sort of simpler models, but I don't necessarily think it has the ability to solve extremely unique problems. So like my time at PlayStation, we would we'd tried a lot of, you know, off the shelf solutions for things like identifying which character was on screen in any particular game.

Andrew: [00:20:39] So we could do some things with that. But but all of the off the shelf models were tuned to things like the real world. So humans and animals and street signs and traffic lights and whatever it may be, you put it into a game where it needs to identify some sort of robot dinosaur. Then it's like, I have no idea. You need somebody in there to be putting the nuance on that to make it actually work. I think for me as well, like Auto, where ML doesn't have any business sense necessarily, so it doesn't know what problems to solve or it doesn't know why it should solve them. So I think humans are still a huge part of that. I don't think that's going away anywhere soon. It's just an evolution and data scientists are going to start, you know, there's going to be bits where automation comes in and helps us do our jobs even better. But I don't think it's going to take away jobs necessarily. I don't have any particular fear about that.

Harpreet: [00:21:26] Yeah. What do you think that's coming from? Like I hear that come out and I share your view on that. I'm like, Yeah, I mean, come on, it's not really going to if you're worried that's going to take your job, you're probably right because you're probably the type of scientist who should.

Andrew: [00:21:38] You post that the other day? I think I saw that. Yeah, but you just said, you know, I'd like.

Harpreet: [00:21:43] To take on that, but yeah. Where do you think that's coming from? And I feel like. I feel like it's like unnecessary fear mongering. And. I don't know, man. This is why I'm going to vote, I think.

Andrew: [00:21:55] I think it comes from social media, right? Social media is this warped view of reality [00:22:00] in everything. And you initially think of that being a problem on Instagram, right, with people and having body dysmorphia. Maybe you don't think about it as being a data science problem, but. Social media is a platform where people share their perfection. Right. And they don't show what's not not going well or the the tape which is holding together their dodgy code and whatever else. So you see things like GPT three coming out, the language model and it can write all of its own code and people go, well, game over for me, right? Like who needs who needs to hire me now? But that's happening at the cutting edge. But we over we overestimate how much of an impact that's going to have on the whole industry. So 99.9% of businesses aren't in a position where they are going to be able to get GPT three to write their code for them. And if it does, they're just going to blindly trust that it's doing the right thing. No, I think the warped view of social media has a really negative effect in a lot of areas of life, and it even creeps into what we do to which is scary when you think data science and data in general, we should be the people who are the most logical. You know, like people with the most common sense thoughts. And we judge everything by its merits and we look at the data and stuff, but still you see some shiny new thing and you go, Well, that's it. So machines are writing code now better than I can write code. So who's going to hire me? But that might be true for. Some major company that wants to put it in place. But most most companies still honestly are trying to get to grips with the data they have. They're trying to understand how they can be more data driven, and they just need somebody to help them do that.

Harpreet: [00:23:57] What's kind of like the thing that you're maybe most [00:24:00] excited about or most hopeful about? Is there is there some application of machine learning, data science, whatever you want to call it in general, that you're kind of excited or that you want to see happen?

Andrew: [00:24:14] I know I'm I'm blown away with the transformers that are coming out. I mean, I'm blown it. I've not worked with Transformers a lot. I've got a book on Transformers that I try and read when I have some spare time, but I'm not that close with how they're working. But I've seen snippets of something which I think is really cool. Obviously, the, the, the models which were performing really well are just beasts. They're immensely big. Whereas now people are trying to be clever and trying to scale them back in terms of. How many parameters are actually in the network, and it's finding clever ways to make it outperform that much bigger model. I think that idea of making these things much more lightweight, that's that's super clever stuff that people are doing there. I'd love to see I'd love to see some more progress within computer vision. I feel like this sort of hit a bit of a ceiling in computer vision a couple of years ago, and there's been no next step that's happened in terms of something amazing. You know, we just know that we can fool a computer vision model with a small square, you know, of some color, and it just completely throws it off. I'd be excited about seeing the next big thing in computer vision.

Harpreet: [00:25:27] I think you work on computer vision stuff at PlayStation or make that up.

Andrew: [00:25:34] Yeah. Yeah. A lot of computer vision stuff for PlayStation. Yeah. My role. The PlayStation was so cool. It was like I was consulting there, but it was kind of an A prototyping team. So it was. It was kind of like the dream role. It was like, come up with cool stuff that we could use as features for the PlayStation five. So there was me and another data scientist. I mean, there was a bunch of engineers from all sorts of different skill sets, and we were just [00:26:00] a team of, of people who tried to use the data that we had to come up with cool stuff. It was like the dream role for a data scientist. I think there was no like specific thing you had to be doing. You had to be trying to be creative and come up with new things that could be mind blowing.

Harpreet: [00:26:19] That's pretty cool, man. Like for working in that type of role. Was that were you working with like data that was tabular data, unstructured data? Like what type of data were you working with? And I mean, obviously without spilling the secret sauce or and NDA violations or anything like that, could you talk a little bit more about that?

Andrew: [00:26:41] Yeah, I probably can't go into too much detail, but yeah, honestly, it was it was a mixture. There was obviously a lot of video data. You know, I did bits and pieces with other guys in the team would do bits and pieces with trying to come up with ways to understand exactly what was happening or overlay things on, on video in real time. And, and, and then there was tabular data as well. So we would have, you know, a PlayStation game, as always, or a game. And General is always emitting that telemetry. You know, exactly what's happening at any point. And like any role, it was a big part of it was dealing with real garbage data. And because the game wasn't the game data, the telemetry wasn't really designed to be used features, it was just created for the developers of the game to kind of keep track of what was going on or ought to be used for other logic in the game. So in a lot of times it was inconsistent game to game because there wasn't a great format. But PlayStation had made some real big strides in terms of making that a lot more consistent while I was there, and that was going to be the biggest win, I think, for them disregarding all of the cool stuff you could build, getting the data into the right format in the first place. So people like me and the data scientists that were there [00:28:00] could actually start working in a streamlined way. That was going to be the biggest one, which is what we always know. Right? We know at the end of the day, the data has to be right for for data scientists to actually do their magic. And that's why data engineering's come, you know, to become such a the new sort of hot topic. It's not because data science was failing, it was because people thought data science could do everything. And that's not never been the case.

Harpreet: [00:28:26] Yeah, telemetry data is really interesting, man. I was that I was at the gym last week. At some point I was doing some some rows had my apple watch on and my Apple Watch just knew just from this motion that I was doing a rowing exercise and know just being a data scientist, machine learning practitioner, trying to deconstruct those was like, how do they how do they do that? Right. Because all you have is a watch, right. And so, you know what what input data would need to would the watch need to know? Probably need to know. I don't know. It's movement along the X plate or its movement along the Y plane and like just over time and yeah. Just thinking about how to reverse engineer the models. I don't know where I'm going with that, but it's a fun exercise to do as a data science.

Andrew: [00:29:08] So I think the PlayStation controller, so when you're using a PlayStation controller, you can control some of your games by moving the controller itself. So it has it has I can't remember the word for it. It's got an accelerometer in it and it's got a something else because. Because it has all of the dimensions, doesn't it? And I worked on a project with some of the other guys there, which was is is actually being patented by Sony. And it was using the, the controller to identify who was holding it so you could have a signature move that you used, which would tell the PlayStation that it was you without having log in or something. It was a bit of a it was one of the sort of leftfield ideas that we had, but it was a super cool project. So using that data, getting people to create a pattern, and then using machine learning and deep learning to understand [00:30:00] exactly whose pattern that was, that was quite a cool mini project.

Harpreet: [00:30:04] Yeah, yeah. I always find that so much fun as, as a data scientist when I, when I see something and I know under the hood that okay, this is happening because of the machine learning algorithm in the background. I was trying to deconstruct what it is that it's doing. What's the input data, data that it just took from this, you know, action that I took. And how did it you know, how did it infer that this would be whatever the right thing to do at that moment? I don't know. Do you ever do that as data scientist? Try to deconstruct, like machine learning in the real world? What's that? What's like? What's like. Something that's just been kind of like tickling your mind lately in terms of in terms of that. I don't know if I don't know of anything recently.

Andrew: [00:30:41] I can't remember the reason I still am. I still am blown away by things that aren't like data science related but are maybe mechanical. So you're sitting at the traffic lights and there's like it was like a garbage truck or something in the way that all of the moving parts were were working together based on some sort of computer system, which has got some logic put into it. And I was like. As somebody coded that up, is that like. Is that some code? It's like not it's not Python. Like it's not some version of Python. It's some, like, hardcore code. I honestly, I don't know how it works to the point where like I think about it as the, like the toaster or the washing machine. And I'm like, how does that work? I don't know. I'd love to know. I'm sure. I'm sure somebody could tell me how it all works. I honestly don't know. But these are these things that they pass you by. Like we're in a world now where all of our music's digital. You know, we've gone past CD-ROMs. We've gone past cassette tapes, like cassette tapes. You know, they were big in, like eighties and nineties. I couldn't tell you how cassette tape worked, but it plays music. It's just like voodoo magic, but that's like 30 year old technology. And I'm just like, I have no idea.

Harpreet: [00:31:55] Yeah. So it's fun to be curious about things like that, man. Like, that's one of the joys I have in life, is trying [00:32:00] to deconstruct explanations for things like, I don't need to know how it works, just explain how it works. And if I understand the explanation and I can connect enough dots, I'm good. I mean, you know, I've got I've got I've got like an 18 month old baby now and just watching the shit that he does and says and how he learns and trying to relate that connect. Okay, here I see natural human learning and here I am as a machine learning practitioner trying to draw parallels between his neural networks and actual neural networks. It's been super fascinating and super fun.

Andrew: [00:32:34] Totally agree. Yeah. Watching kids learn when you're when you have done deep learning is fascinating. Like the idea that they can identify like they might see a picture of a lion on the TV a couple of times, but then they'll see a cartoon lion, which doesn't really look like a lion that they've seen, but they'll go, There's a lion. And you know that no deep learning algorithm would have figured that out was lion no chance. So you're like, What is going differently? What is going on differently in their brains? And that's what I'd like to see this this next step in computer vision. I know Geoffrey Hinton talked about capsule networks a few years ago and I've not seen anything more on that. That would be cool. But yeah, who knows? There must be something big coming up, surely.

Harpreet: [00:33:17] Yeah, that's fascinating stuff. But I just intelligence in general, whether it's artificial machine learning or human intelligence, it's just so fascinating. Man, before we get too philosophical here, let's talk about talk about your mission to help develop data scientists. I mean, obviously, I think it's a mission because of data science, infinity and the work you do there, and then all the sorts of webinars and workshops and stuff that you do. But where did that desire come from? How did that start?

Andrew: [00:33:44] Well, I I've always had the opportunity in the roles that I've been in throughout my career, from time to time, to sort of mentor the junior analysts or, you know, whoever I may be enough was really enjoyed it. And I'm not a super extroverted person or anything [00:34:00] who necessarily goes looking for that. But, you know, the opportunities that I've had, I've found it really fulfilling. I even vividly remember one time helping a junior analyst who really wanted to get into sort of machine learning. But it was sort of stuck in a team that was just doing dashboards and analytics, and I managed to help her. Buddy up with somebody in the Advanced Analytics Team where I was working, work on a machine learning project, and she did really well and I think she ended up moving into the Advanced Analytics Team and it was like I felt like I had made this sort of difference in somebody's life in the world of data science. And I vividly remember that because it felt so good to have helped somebody figure out what they wanted to do, but also help them get there when when nobody else was willing to do it. So that said, I've always found that part of. My career fulfilling. And then in terms of data science, infinity itself, where that came from. So I had been consulting at PlayStation for about three years, I think, and then that just happened to come to an end because the tax rules in the UK all changed, meaning that the way that I was consulting and the way my business was consulting with PlayStation, it all changed.

Andrew: [00:35:11] And then that sort of whole contract consulting, consulting market, just the bottom just fell out of it. So I, I'd been thinking about doing something. On my own in some shape or form, whether that was like a consultancy or whether it was teaching or whatever. I'd been thinking about the idea of doing something for my own company for a while, and I had all of these useful code snippets and all of these other bits of information that would be great for a for a course. And I also realized that I had some experience that not many other people had and that I had been very fortunate in my time at Amazon and Sony and even before that, to have interviewed and screen, you know, hundreds and hundreds of candidates. And you just don't really see that that often. I was just been very fortunate to have been put in that position and I enjoyed [00:36:00] doing that. So I always sort of tried to try to be in that position as often as possible. And not not only was that experience of interviewing very useful, but I had just organically built this very sort of acute view of what it was that discerned candidates who got the role versus those who just missed out time and time again. So I kind of at that point figured that there was something I could use for the content. And then there was at least a point of difference for me to try and say, Well, here's something that could be valuable to you.

Andrew: [00:36:31] So I took the plunge and started working on what would become data science infinity, and I didn't want it to be based on my opinion. During all of those interviews. I'd seen so many people come in with the certificate and that certificate and they just it just didn't. There was a disconnect between what they were coming in with and what hiring managers or interviewers actually needed or actually wanted. So I didn't want any of the content or any of the sort of advice to be just my opinion, because I think that's a big mistake. I mean, you'd never. You'd never like a data science or or a statistician who relies on a single opinion is almost it's it's ironic almost, isn't it, that using a sample of one and you'd never use a sample of one in your work. So I went out and I thought, I'll go talk to a bunch of other people. So hiring managers, leaders and recruiters and I went and talked to hundreds of them and I still talk to hundreds of them. I've got my sort of group of people and I can continually reach out and try and update that view of the industry. So I asked them all about things like skills and tools and education and interviewing. What's the difference between great data scientists and good data scientists? To add to my own experience, what's the difference? What is it that differentiates people who manned roles versus those who miss out? Long story, short boiled boiled.

Andrew: [00:37:47] All of that down to the initial data science infinity curriculum. So data science infinity now is basically around learning the right skills that hiring managers actually want, and that's come from hiring managers mouth rather than [00:38:00] a job description, which we know of. And 80% of the time it's just garbage. It's all about learning in the right way with a focus on intuition and actual hands on application and then actually landing a role at the end of it. Because 99% of the time that's what people that's their goal. They don't just want to learn it because they're fascinated by it. They they want to move into it. They want to change careers because they're unfulfilled with their career or maybe their industry is being automated and they want to move to something which is a bit more futureproof. So I thought, you know, I at least have this. Point of difference. So I went for it and it took me about seven months to build up all of the content, which was five months more than I thought it would. And then like I was saying to you just before we went online, I just realized the other day that data science infinite is being live for a year. Last week. I think it's been a massive learning curve, like the data science stuff itself. That's kind of fine. I'm happy enough with that. But marketing and selling and pricing and like I didn't know how to do that. The first year has been a massive learning curve different.

Harpreet: [00:39:01] And I can imagine as a as someone who's launching my own course, there's a lot to learn, but I do. Data science affinity is definitely an excellent program. I've signed up for four for that myself. So seeing the work that you put into it, it's definitely good stuff. So speaking of the secret traits of great data scientists or secret traits, so those who get hired share a few of those with the audience here. What what makes a an employable data scientist different from a unemployable one?

Andrew: [00:39:33] Well, I think at the end of the day, I think the difference between a good data scientist and there's a lot of good data scientists, and then there's the great data scientists. And I think the the difference between good and great is often the soft skills. And we talk about this a lot. I know you have your views on it and it's it's fairly aligned with with mine, I think. And in my experience and this is something I say all the time, that by far [00:40:00] the best data scientists that I've worked with in my career at companies like Amazon, where the people are incredibly talented, they're not the smartest people by definition. They're not the best mathematician, the best coder. Absolutely. Absolutely. They know their stuff in terms of their coding and their statistics and their other key data concepts and tools. But they are the great ones are the ones that can understand what the business is trying to achieve and then use data and their skill set and clever and often quite simple ways to actually add value to the business or to the team. I think communication is a massive thing that differentiates the majority from the top tier data scientist. I think you could summarize it down to say that a good data scientist knows a lot of technical concepts, but a great data scientist can simplify those right back in a way that everybody in the business can get on board with whatever the project is, no matter how complex it is.

Andrew: [00:41:00] And at the end of the day, I think great data scientists have this. They're able to view what it is that why are we here? They know that they're there to actually solve problems, not create new problems. Finding problems, finding difficulties of why something won't work. Let's find a way that it will work. Whether that's simple or whether that's complex, let's find a way to make it work and let's do it. And they know that they're there to enhance or accelerate business decision making. Again, you know, same sort of thing, not get in the way of it. They're there to add value. At the end of the day, you're being paid. So there's an ROI. People are expecting you and what you do to have a return on investment. So you're sitting in your chair being the best mathematician in the world or the best coder in the world. But you can't explain anything to people in a way where they get on board and whatever it is you're building actually gets implemented and starts adding business value. Then what's the point of view being there? You're great, you're [00:42:00] smart, that's awesome. But you're being paid for a return on investment.

Harpreet: [00:42:06] So I guess where do you think most data scientists go wrong in terms of their own career development?

Andrew: [00:42:14] As far as to simplify it down in terms of career development and trying to land roles, whether that's your first role in the industry or maybe you're looking to move from your role up to a bigger and bigger role, I would say, and this is something that I see both on resumes, the way resumes are written and in interviews as well is. Is this the first selling point they see for themselves as slightly off? Only by a bit. But it's they go in with the sales pitch of look at what I know how to use. Here's the list of things. And it's not the best way to sell yourself to somebody who is. Who is they hiring? Somebody. Not so fun. They're hiring somebody because they've got a problem they need to solve. Or they're so busy that their team can't keep up with the work, they need somebody to come in and solve problems. So instead of this narrative of, Look at what I know how to use. It just needs to be tweaked slightly to be, look at what I can do with what I know how to use and that simple change in terms of your mindset. And then it just flows onto your resume and the way that you describe your projects and interviews. That is going to make the biggest difference at the end of the day. So if you've got a data scientist coming in for an interview and they're saying, I built X and it drove Y million incremental dollars, or I built X and at say Y analyst hours or I built X and I reduced churn by 1%.

Andrew: [00:43:44] That's a I'm simplifying that write down. But it's not just I used Python, I used TensorFlow to build this cool thing. It's I built something and look at what it did. And then all of a sudden the interview is going, this person that's talking to me, the [00:44:00] cogs in their mind are churning the saying, this person's going to either solve the awful problems that I'm struggling with, either this person is going to make me and my team look really good, which is, at the end of the day, a big part of what people are looking for or. This person that's talking to me, they're going to make the business money and make all of the the beer side. That's what interview is looking for. They're not looking for just another person who knows place. They're looking for somebody who's going to solve their problems and going to add value, you know, using those skills. Don't get me wrong, you need those skills, but you want to sell yourself based on the value using the skills, not just giving a list of things that you know how to do.

Harpreet: [00:44:42] Andrew Let's take some questions coming in from LinkedIn, a bunch of questions coming in. I think we could have some of these people happy. By the way, on LinkedIn you're watching everybody smashes smash. Your reaction real quick, man, and share this with the network. I see I see a bunch of you watching, but not enough reactions and not enough sharing with the network, my friends. All right. So let me Chandra has a question, general question. But how to choose the right model to train the data? It's one of those very questions. You give it a try. Should I give one of my standard responses? Yeah, I.

Andrew: [00:45:19] Think I think there's probably another couple of questions before that. But, but I wouldn't think I wouldn't think what's the time model to train the to use on the data. Again I would think what is the what is the end result we're looking for? What are we trying to solve here? And so something I say all the time, harpreet, I'll let you answer as well, because I know you'll have a good answer for this, but something I say all the time to my students and data science. Infinity is always, always, always start with the business problem. Get a really good understanding of what that is, what success might look like, and then work backwards from there to a data science solution. Don't do it the other way around. Don't start with the data science solution, then try and force that into the business problem because you never align [00:46:00] the two that way. So start with the business problem and then work back.

Harpreet: [00:46:04] Yeah. It's pretty much say the same thing like I don't think it's ever really. The first question to ask is how do I choose the right model to create it? I mean, okay, let's say you've done everything up until that point and now you're like, okay, I'm ready to, you know, get a model to this and figure something out. How do I move towards the right one? Experimentation that is the science and data science, right? So maybe further upstream, you plan out in advance that because of the business problem, because the problem statement, because the nature of the data, because of what it is that I'm trying to predict, given all this background context, here is the suite of algorithms which I'm going to try. To solve for this problem and of this suite of algorithms. Maybe I'll do some statistical tests to see which one truly is performing better. Obviously, I'm assuming you've already got a baseline in place, right? Because you have a baseline before you start more complex things. Right? So this is the whole process. So I would say first, you don't know what the right model is going to be ever. You start with baseline, start some candidate algorithms, see which one of those candidates perform best and then optimize the best performing one. There is no right, right model.

Andrew: [00:47:21] And then on top of that, exactly right. On top of that, you've also got considerations around what's needed surrounding the model. So you say very simply, if you were to compare like linear regression and decision tree and random forest, for example, like a random forest is a is a little bit harder to interpret exactly what's driving the predictions. The decision trees super easy to see that linear regression can be really really useful for that so so it kind of it's not all about model performances. It's sometimes there's other things that need. Sometimes you don't need to know what's driving the model. You just need to get the best prediction. Sometimes you [00:48:00] need to be very well aware of what it is. If it's a if it's a churn model, for example, knowing exactly what the drivers of churn are is extremely important because then the business can actually action something off the back. It And the other thing I'd add to that as well as I'm a massive fan of starting with an MVP mindset. So MVP being minimum viable product for anyone that doesn't know that acronym. But and that's not saying that you definitely have to go and build an MVP, but but in your head, think about what's the what's the absolute bare bones solution that would. Get us to a place where everyone in the business that needed to be involved could understand what we're going to do and what what the output is going to be and where the touch points are from different teams. And then you can go and then build a version two, which has all the complexity in the world, but start with a simple solution that people can buy into.

Harpreet: [00:48:53] Another question coming in here from from Christine Seagrave, a good friend of the show, how can machine learning develop a means to incorporate underlying human emotions that underpin decision making? Is there a field within machine learning that focuses on incorporating human concerns through technology development?

Andrew: [00:49:12] Interesting question. I don't I don't have a super good answer there. I guess I guess what the advances in language modeling at the moment is. You could you could infer emotion from what people are saying. I know from like more of a computer vision approach. There's been a lot of attempts to understand people's facial expressions and things like that, but I don't think they've been extremely reliable to the point where they were used in an interview scenarios, for example. And there's a lot of not only is it has to be very performed very accurately, but also the sort of ethical areas around that as well. Yeah, I don't know. Human emotion is not something that I've necessarily [00:50:00] seen being as the number one focus of a deep learning model, for example.

Harpreet: [00:50:05] Second question, I think that would be ethics is a field within machine learning that focuses on incorporating human concerns into technology development. I think that would just be ethics, machine learning, ethics, something like that, something that I find interesting. But I don't know of many resources that discuss that too much. I mean, there's a few books Hello World by Hannah Fry, I think touches on that. Weapons of Mass Destruction. There's books on the topic, right? But not like textbooks or anything like that. Definitely no boot camps that talk about it, that's for sure, which they should be. Great question there. Russell Willis, good friend of the show, says accelerometer and gyro data, I suspect, I guess that was in respect to the Apple Watch being able to predict which activity I'm doing. But even just think about the entire like how they made that happen, they probably had to pay a bunch of people to wear a prototype, Apple Watch, have them go exercise. And then you probably had to have a whole range of people from healthy and not so healthy and fit. Not so fit. And then think of all that training data they probably collected just and you remember, they can only collect data that's here.

Harpreet: [00:51:15] So they probably had heart rate data and like you mentioned, accelerometer gyro data, they probably had something like baseline heart rate to accelerate in heart rate after some minute of time and things like that. And you think, okay, great. They paid a bunch of people, collected all that data and then they took all that data and then they probably, I don't know what type of model they used, maybe the rand or something like that, and then take the AR and whatever ensemble of different models they have and deploy it on edge device that then just takes essentially heart rate accelerometer and gyro data. And it's just mind boggling to me. That's the kind of stuff I like doing. I like reverse engineering, machine learning. Um, literally, I've been thinking about this for, for the last few days. So Christine wants to know, what advice do you give social scientists [00:52:00] that are learning data analytics? Any particular hints for psychologists trying to understand acceptable norms of behavior when creating data science projects?

Andrew: [00:52:11] Oh, I don't know. I don't know if I've had any specific conversations of that that type. I would say in general, where I've found my this is not kind of exactly answering your question, but I found that psychologists moving towards data science or data analytics is actually a really good skill to have. And I found it really, really handy in my career because I guess psychology at the end of the day, depending on which part of psychology you're in, but it's very much concerned with why people do things. And at the end of the day in data science is kind of what we're focusing on too, right? If you're in a retail business and you're working in data science, a lot of your work is thinking about why did people take this action? So having that curiosity around that is a very good foundational skill set to have for a data scientist. I know that's not exactly your question, but I didn't have a good answer for your question, so I just went off on a tangent.

Harpreet: [00:53:05] Yeah, that's a great question, Christine. I mean, social scientist, you're probably working with data, you're probably working with observational data. And let's face it, most of analytics that you're doing in a company as a data analyst, it's pretty much observational data, right? Unless you're out there designing experiments yourself, which that's also a social scientist do. So I think there's a lot of overlap between social scientists and then data analytics and data science. So any particular hints for psychologists trying to understand? Digital norms of behavior when creating data science projects. Yeah, be curious. I think that is probably the the most acceptable norm of behavior is be curious, be fun and just have fun with it. So. Couple more questions then we'll get into a real quick what I like to call a random round here. So so you got this awesome YouTube video if you guys haven't already, check out Andrew Jones Data Science Infinity on YouTube. He's got some really fun, interesting videos, [00:54:00] really well produced and some awesome thumbnails as well. But you got one that talks about the top five reasons that candidates get rejected. I was wondering, you can walk us through those those reasons.

Andrew: [00:54:11] Yeah. That was a that was a we wanted to go now. I think I kind of know because the sort of the five that I boiled down from one of the data that for me talking to recruiters and hiring managers, I think I think the I think the five reasons why people were rejected from from interviews and that seems quite harsh. Maybe the reasons that they didn't get the role this was other candidates maybe, or they weren't ready for the role. So number one, I think very, very basic. You just don't have the right skill set. So as much as I talk around, it's not all about just knowing these things, it's about the value you can add there and skills that people will require you to have to be able to work with their infrastructure. But you need to be able to work, work in a way with those skills that can be efficient and effective and solving business or customer problems. I think another thing is candidates maybe being in this relates to what I just said, candidates being too focused on tools and concepts and not focused enough. And this is not only in the way that you work in general, but the way that you convey your ideas to an interviewer too, focused on tools and and concepts, not enough about understanding business problems and talking about ways that you can solve problems and the impact that that might have on the business or on the customer. I think seeing what else was there, I think something that people mentioned a lot was candidates who have obviously jumped ahead to topics like deep learning without necessarily putting in the groundwork with the more classical approaches.

Andrew: [00:55:47] You know, this is sort of machine learning in general. That's only one part of data science anyway. But people that have obviously jumped ahead to deep learning without that really good understanding of what the classical machine learning algorithms can do because they're amazing. [00:56:00] They can do they can solve 95% of the problems that need a machine learning solution anyway for most businesses. So so doing the groundwork, putting in the hard yards to understand how they work in a way that means that you can manipulate them in different ways to solve different problems. So a logistic logistic regression model is not just there to. Classify churn, churn customers or whatever it may be. There's so many other parts of how that works and how it can be used. So if you have a good understanding of that, you can be really creative with it. In terms of your approach to the interview itself or applying for the role, I think in terms of your portfolio of projects, this is more for people who are early in their career and you've got a suite of projects that you want to showcase to people. They've gone for projects which are all about complexity. They've tried to do the most complex thing in the world to show that they're good, but that is often not the way to do it.

Andrew: [00:56:59] It becomes very hard for a hiring manager or created to understand what you're doing. So. So instead of complexity, I always say try and go for portfolio portfolio projects that are have a much higher emphasis on clarity and impact rather than just complexity for complexity sake. And then the last one I think is it's quite a broad area, but just this idea of communication. So and I've talked about this a couple of times today as well. It's it's not just being you can't just be clever. You've you've got to be able to work with other people in the business who are from all sorts of different backgrounds, of all sorts of different skill sets and the stakeholders and the management and the business. They're the ones who who. They will be the ones who give the green light for whatever it is that you've built to go into production or to to go on and impact customers. And if they don't give you that green light because they don't understand what it is that you've done, because you've not been able to explain it to them, then everything that you've worked on just sits on the shelf, adding no value. [00:58:00] And if you don't add any value, then you're not going to move up in your career whether that's getting promoted internally or whether that's moving to a new role, because that is what people need to see. They need to see evidence that you've used your skillset to add value.

Harpreet: [00:58:13] When it comes to career growth and development. What's the biggest lesson you learned the hard way that you want to make sure no one else makes?

Andrew: [00:58:23] I think something that I've been guilty of earlier in my career is this idea. And I talked about this before of. Not starting with the business problem and working back to a solution. So there's some awesome stuff that you can do in data science and you get a project that might need some sort of classification done. And you go, Do you know what I was reading about this awesome deep learning model the other day, this new architecture that I saw, and I'm just going to try and force that in and kind of convince people that we need to do this because it's fancy and. That never ends that well. Sometimes you can get lucky and it will be fine, but often because you're trying to force something in, it doesn't necessarily align with what the business needs or in terms of the actual outcomes or the timeframes or the it doesn't work with the architecture that they have, or maybe the people in the business won't understand it as easily. So starting with the business problem, getting an understanding of understanding what success might look like, understanding who needs to be involved. Get that down first and then then get your data science hat on and start working. And that sometimes that's a less exciting way to do it. But if you want to be a value adding data scientist, it's the best way to do it. And I've been guilty of not doing that in the past. I would say.

Harpreet: [00:59:44] The last formal question before going into I like to call the random round. It's 100 years in the future. What do you want to be remembered?

Andrew: [00:59:53] I would say if I'm serious, it would it would be nothing to do with data science. It would [01:00:00] just be being a good dad and a good husband. My my family is my whole world. So that would be the big one, I guess. Somebody who doesn't I'm trying to think about the traits that I would like to be remembered for. I guess somebody that that doesn't necessarily just blindly go and do what everybody else does because it's the popular thing to do. And somebody who stands up for what I believe in, I think there's a little bit of a lack of that in the world right now, being somebody who's somebody who maybe supports other people rather than tries to bring them down, limit them because you fear that maybe they might become better than I would. You know, I think that's a big part of what I what I love about data science. Infinity is trying to help people get into the industry that I love because I don't see any. I don't see that it's a closed industry and getting more people in is going to impede me in any way. I think. I think the more people we get in and the more people that can be working on the types of problems that we have, the better. And that's why data science, like I say, is so rewarding for me. Anything I can do in that sense is something that I'd like to be remembered for, I guess, because it makes me happy. I suppose so. I guess it's something I should keep doing.

Harpreet: [01:01:05] Absolutely. Andrew, thank you. Definitely will be remembered for for all that man doing awesome work. And I know you're a great family man as well. So let's jump into the random round. First question I want to ask is, what do you think will be the first video to hit 1 trillion views on YouTube? And what will that video be about?

Andrew: [01:01:25] It would have to be something like like a K-Pop song or something, something that's just massively popular, which is just slightly bypassing me at the moment, because sometimes I go and Google that, like Google what's the highest viewed video on YouTube and like six out of the top ten, I'm like, I've never heard of that. Whereas some of them are the classic, the classic ones, anything. The thing that makes me lot like people like those fail me videos where people are just getting destroyed, like falling off motorbikes. I think that's the best stuff ever, but I'd love to see one of those big [01:02:00] I don't know. What do you think, Harpreet? I'd love to hear another opinion on that.

Harpreet: [01:02:03] Yeah, this is this is something I've been asking a lot of people. You know, it's one of the questions I asked in my random rant most frequently just to, you know, wisdom of the crowds thing, see what happens. And based on what people have told me, it's likely going to be video involving a cute baby or a cat, and it's probably going to happen by the year 2030. That's kind of been.

Andrew: [01:02:24] The what's the highest views at the moment? It must be a couple of like 100 billion or something.

Harpreet: [01:02:29] What would be shark with like 9 billion or 10 billion views? So it's still a long ways off, baby shark like the high speed, I think. And that just like in November 2020. It passed up like Despacito, which had like nine or 10 billion views. Like that 9 billion views. So, I mean. And I think the first view of Crack 103, the first video to crack 1 billion views was Gangnam Style back in.

Andrew: [01:03:02] I know Justin Bieber's got a couple that are up there, doesn't he? He's done very well on YouTube. I love the baby shower. It's the top one. Not that I love baby shower explicitly because I've heard it 100,000 times myself. But I love that there's somebody who who wrote a song called Baby Shower, and I hope that they're just swimming in cash for something like that. I'd love I'd love to know that they were living in a gold mansion because of that.

Harpreet: [01:03:29] So what are you currently reading?

Andrew: [01:03:31] Oh, I don't. I don't, actually. Well, I. I don't do a lot of reading for pleasure. Like, I don't read novels or anything. I just have a my office is just full of like you can see in the background there. There's some of the some of the books that I'm not reading at the moment. It's just all deep learning, machine learning data science books. And I don't I don't like to read them cover to cover. I never do. They just like reference books. And I have this Oscar up. I was going to [01:04:00] do a video or a post on this because I'm curious about if other people learn like ideas or function or learn something. I'll have three books all open on on on sites. Deep learning something and deep learning. Yeah. And I'm trying to figure out a way to learn it myself. I'll have three books open to that particular chapter and I'll be reading all of the chapters at the same time. And it's basically like having three people talking to me about the same subject, and I can pick and choose what makes sense to me and what doesn't, and I can sort of almost piece the puzzle together. I don't know if there's a name for this type of learning. Yes, but I do it all the time.

Harpreet: [01:04:33] For that type of reading in particular, it's called optical reading, SWI and Optical Optical Reading. Yeah, I did that as well when I was going through some deep learning. I mean, I'm still learning, deep learning, but I had three books in particular, The Visual Introduction to Deep Learning, which was the book by Andrew Glassner. I had John Chrome's book Deep Illustrated and then Andrew try and yeah, I do that as well. Yeah.

Andrew: [01:04:59] Well, good and good. Knowing the name of what that was called, you sound like the sort of person who'd be good on a pub quiz team.

Harpreet: [01:05:05] Probably that maybe limiting trivia for a while. So what song do you currently have on repeat?

Andrew: [01:05:12] No baby shark at the moment. We've passed Baby Shark where I'm currently. I don't get to choose a lot of the music in my household. We've got trolls, world tour songs, so trolls just want to have fun, which I actually like. That's really cool. My daughters love the song Roar by Katy Perry, so dance on repeat at the moment and then contrasting to those is their Metallica's Black Album came out. 30 years ago. The other way. And so they've they they released this kind of album of people doing covers of their songs and about 20 different people covered. Nothing else matters, because it's one of the versions I don't know if you've heard it is by Miley Cyrus and Elton John. So Elton John's playing piano [01:06:00] and Miley Cyrus is singing. Chad Smith from the Chili Peppers is on drums. And it is awesome. So that's like the only song of my own choice, which I'm on have gone on repeat at the moment. So trolls Katy Perry and Metallica, an unlikely trio.

Harpreet: [01:06:17] A huge chili pepper. Sounds like a favorite band. Check that out. We're going to go to the the random question generator and will be a lot of fun. All right. First question, the random question generator is, if you had to change your name, what would you change it to?

Andrew: [01:06:33] So so my last name is Jones. My my wife didn't change your name when we got married. So her last name is still her last name, which is Maori. But we've always debated both changing our last name instead of her being Murray and me being Jones to just mujer. But like genuinely just just both changing names because because in general, the tradition is that the the the wife changes her name and she just has to kind of do it my my way said now I'm not going to do that. And I was like, That's cool. And then we thought, What if we both changed our name to just mujer? I think it would be super fun, but I'm not brave enough to do it like that.

Harpreet: [01:07:10] Mucho. That's cool. Yeah. I mean, like, I don't really care about that either. Like, my wife didn't change her last name. I was like, Dad, I don't really care if you do that or not. Then my son has the hyphenated last name.

Andrew: [01:07:21] Just same as.

Harpreet: [01:07:22] Us. Yeah. We are. I like that. What's on your bucket list this year?

Andrew: [01:07:30] Oh, man. Bucket list items over the past 18 months have been a bit tricky, haven't they? My. The what the place in the world that I love the most is is Dubai. I just love going there and staying in a nice hotel and just getting the lake 40 degree Celsius, 40 degree weather and just relaxing there. I love it so much. I just want to go back so badly. But at the moment, with travel being a bit tricky, it's just going to have to wait. I [01:08:00] want to go back.

Harpreet: [01:08:01] I've never been to it. To Dubai. That's on the list of places that I want to go. I'd love to check that place out, but at 40 degrees Celsius like that, I don't like hot weather. I just.

Andrew: [01:08:11] I do. I do. I do. If I'm, like at a hotel by a pool, if I'm like in the city of London or something, I'm not interested, you know, like walking down the street for work or something. Not interested. But yeah, if you're by a pool, I can deal with some some hot temperatures.

Harpreet: [01:08:26] Yeah, man. I think for me, perfect weather, probably 21 to 22 Celsius. I think that's like absolute perfect. What's that in Fahrenheit? I don't know, 22 C to F that is 71.6 degrees. What's the story behind one of your scars?

Andrew: [01:08:42] I've got a scar that runs just across here on my forehead. It's fighting a little bit now. But I went to I don't know how many years ago now, maybe like nine years ago I met a bunch of my mates from New Zealand and Thailand and we went to like the full moon party on, on New Year's Eve and there was big flaming hoops. And in Thailand you drink from a bucket, which is basically like a Red Bull and vodka. But the Red Bulls, not Red Bull. It's like this, possibly some sort of gasoline in there. And. Long story short, I thought. I can jump through that, hoop that super easy jump through the hoop. I actually made it through the hoop. But when I went to do my landing, I cushioned my fall with my own knee. And so just like split open my head. But I mean, it didn't dampen the night. Obviously, I'd had enough Red Bull and vodka at that point to just carry on. So that's my scar story.

Harpreet: [01:09:42] But that Red Bull that in Thailand, that's like the original, original Red Bull. That's tha that's true. That's the powerful. What languages do you speak?

Andrew: [01:09:52] I speak English, and the only other language which I would say I speak more than a little bit, would [01:10:00] be moldy, which is the language from New Zealand. But the people of New Zealand and I don't speak it fluently, but I know quite a bit of it because I did it all through school and my wife's family are part Maudie, so they're very proud of the heritage and we learn a lot of it to teach the girls as well. So that would be kind of my second language, I guess.

Harpreet: [01:10:22] That's awesome, man. What are you, a natural?

Andrew: [01:10:29] Definitely nothing to do with data science. I have to work really hard for everything in data science, I'd say. I've always had a natural tendency for sport. You know, thing with the thing with a bowl. I just all day every day. I wasn't inside on TV and I think all day, every day as a kid, I was outside kicking the ball around or throwing a cricket ball or whatever. That stuff's my most natural talent, I think.

Harpreet: [01:10:58] I love him. I can see that. I can see that as well. So. Man. It's been an absolute pleasure, Andrew. Thank you so much for taking time to be on the show. How can people connect with you? Where can they find you online?

Andrew: [01:11:08] Yeah. Firstly, my my pleasure. Completely. Absolutely. Appreciate it. So, so good to come on and talk to you one on one finally. I love it. So I'm on I'm on LinkedIn. That's my main social media so you can find me. My name is Andrew Jones and in data science infinity. So it makes a little bit easier because Andrew Jones is an extremely common name. I like you mentioned, I am on YouTube just under Andrew Jones. It's something I'm starting to try to build up a little bit. And then I'm recently just joined Instagram as well so you can find me on there as well. And that's just under data science infinity. But if you connect with me on on LinkedIn and you've got any questions about data science or the transition to data science, then just just message me and I'll definitely get back to you. And then, you know, I've talked a little bit about data science and, and what it is and why it works the way it does. So if you want to learn more about that, then, then the site for that is data science data, infinity dotcom. [01:12:00] So head there. There's all the information, there's the full curriculum which you can use for your own sort of study pathway to towards data science. And remember, that's all based on hiring managers inputs. There's all sorts of preview videos, there's feedback from students, and then you can contact me through there as well.

Harpreet: [01:12:16] I'll be sure to link to that in the show notes link to totally mention that places Andrew thanks again for for taking time and coming by I know I know it gets late for you. I think it's pretty late right now. So it becomes tough to make it to our data science. Happy Hours. But one of these days my swing back to my earlier sessions that I had earlier in the day that I would love to love to have you there as well. Thank you again for coming on the show and I appreciate you being here.

Andrew: [01:12:40] Yeah, my pleasure. Completely. And I'll definitely try and get on to the office hours because like you said, I've got my own office now in the house. So if it's mid-afternoon here, which I think the comment was one was today, I should definitely get it.

Harpreet: [01:12:51] Yeah, absolutely. Thanks again. And my friends, remember, you've got one life on this planet. Why not try to do something big? Cheers, everyone.