#259: From Academia to Tech Industry and Python Transcript

00:00 Did you come to Python from the academic side of the world? Maybe you got into working with code for research or lab work, and you found that you like coding more than your first field of study. Whatever the reason, many people make the transition from the academic world over to tech and industry. On this episode, you'll meet three women who have made this transition and you'll hear their stories. I'm excited to speak with Jennifer Stark, Kaylee Haynes and as lean becoming about their journey into the tech field. This is talk Python to me, Episode 259, recorded April 6 2020.

00:45 Welcome to talk Python to me, a weekly podcast on Python, the language, the libraries, the ecosystem, and the personalities. This is your host, Michael Kennedy. Follow me on Twitter where I'm at m Kennedy. Keep up with the show and listen to past episodes at talk python.fm and follow the show on Twitter via at talk Python. This episode is sponsored by data, dog and linode. Please check out what they're offering during their segments. It really helps support the show. Kaley esslyn. Jennifer, welcome to talk Python to me,

01:13 thank you very much for asking me.

01:16 It's really great to have you all here. I was in a Ph. D. program. My wife is a professor. So I've spent a lot of time in the academic world. And I saw that you all were participating in this, I think at the time was going to be a live event around sort of sharing stories, going from academia into more the tech side of the world. And I thought, Oh, what a fun topic. I have to reach out to you and have you on the show. So I'm happy you're here. It's great.

01:43 Great. Thanks very much. No,

01:44 yeah, absolutely. It's really interesting story, how people spend so much time working on getting a PhD or getting some academic position. And it's just I find the whole academic space is really different than the tech space. And just to give folks a sense, you know, like the story for my wife, if she wants to change jobs, like there's only a certain time of the year that jobs are kind of offered, right, like interviews are done in January, and you apply to them. And there's like 30 or 40 in like a very specific area in the world. And it's just such a different world in the tech space. But at the same time, college campuses are beautiful places like the energy is just I love being on campuses and whatnot. So it's just such a trade off to make I think so I'm really interested to dig into that with you. But before we get specifically that I want to start with your story. How did you get into programming and Python? Jennifer, your first Yeah,

02:39 well as baxton. So I was working in the US at the time in Baltimore, for the NIH, and it was during the recession, I guess so 2009 2010. So the funding was frozen for a number of years. So we had to cut back what we were like the fancy not it wasn't even that fancy. But software, we're using a new open source software. So we started with our so I don't know,

03:05 okay, nice. What we're using before was like MATLAB or SAS or something like that.

03:09 Yeah, some MATLAB and some other statistical programs that were made back there. But some of those were also open source specifically for the kind of data that we're analyzing

03:19 and this I'm guessing is in the health and medical research area if you're in age

03:23 Yes, it was um neider to Nationals cheap for drug abuse.

03:26 Yeah. So you learned our and then somehow you got from our are you doing are only are you doing are in Python. Now

03:33 currently, I've not I've not used offer a long time I moved over to Python. And because the kinds of things I wanted to do broadened and pythons good to so many things. So it just seemed like a natural step for me.

03:45 Yeah, I think that's a trade off. A lot of people have to make between R and Python. I feel like R is very specialized. And it's very powerful. But then Python allows you to say like, I have the same skills. Oh, and I could go build this web app. Or I could go build this other thing that is not actually just about visualizing and processing data.

04:01 Yeah. I mean, I think the last package I used was for looking at diffusion. And the only place you could find that could allow you to do catalytic diffusion was in our package. So like, that's how specialist it is. It's amazing.

04:17 Yeah. Wow. Very cool. As lean

04:19 Yeah. So I started, I think, well, I learned Python when I was in my second year as an undergrad. So in my first year, I was using MATLAB. And my second year, I was introduced to Python. So I was studying math. So yeah, I do a degree in math. And I just fell in love with Python. So yeah, since then, I've just been using Python. But yeah, it was good.

04:45 Yeah, very good. And I also was working on in math for a long time and it's a fun world, but I gotta say, I like programming better.

04:54 Kaylee, how about you?

04:55 Yes, I went to university to study physics and then after a few months, Realize that they really like physics. So I moved into a maths and statistics degree. So I've got the statistics background. So I've come from R. So I think I must alone are about my third or fourth year undergrad. And then from that, I then decided to go and do a masters and a PhD in statistics, I sort of got quite a strong art background. And then just recently, I guess I'm trying to learn Python, but very, very, like beginner level. And I guess I mean, Jennifer sort of knew each other through like pi beta and our Manchester

05:27 nice. And I guess it's worth mentioning that you all come from the same town in Manchester in the UK. So you, you could actually see each other at meetups and whatnot. Right? not just necessarily online

05:37 in the old days. Yeah.

05:40 Yeah, that's super cool.

05:41 Yeah, we're all in Manchester now. But we're not all when none of us are from Manchester or anything. No.

05:46 Well, and you know, I mean, we'll get into it later in the show a little bit, maybe. But the idea of getting together and seeing each other right now as kind of you could be on the other side of the world. It's it's still weird time. So yeah, well, I guess, you know, I hear a little bit of your story and how you got into these these various areas. But what got you interested in sort of computational side of, of computers and programming and whatnot. I mean, Kaley go in reverse order, I guess. So if you're not into physics, and you want to switch into math, like when I was in math, it was pen paper or chalk. chalkboard, right? Like, let there be a theorem, I did end up doing a bunch of cool, like visualization work and stuff do near the end. But it definitely didn't start out that way it started out with, okay, we've got, you know, seven axioms and a corollary, and you've got an hour to come up with. Right, right. So what got you sort of going down that path.

06:42 So I think the reason I changed from physics to maths was I didn't really like the practical lectures at a time that has been taught when I went to university, which is sort of interesting, though, because I'm more interested in the practical side, I think, first year physics when I went to uni was very basic and very boring. And I remember I always like go back to the same lab that we had, where we just had to swing a pendulum for an hour. I'm sure that's in class, but for some people just really didn't interest me at all. So then moved into maths and statistics, I just had this really great lecturer in statistics. And first year, he was so enthusiastic about data, which I think that sort of helped seeing someone else be so enthusiastic about it. So I decided that statistics was the degree that I was going to do. And I think that like my third to fourth year, so we did four year undergrad up in Scotland. So I think in between the summer of my third and fourth year, and then decided to go to an internship at Lancaster University. And there's a really good training center on statistics and operational research. And that was really interesting. It was sort of like an eight week program, where you got given a stats project to do kind of give you an insight into doing a PhD. And the doctorate Training Center. Lancaster is sort of built up and geared towards giving you an insight to do statistics, but in industry. So it gives us a real like practical sort of use to data. So as you said, like a lot of maths was like theorems and just different like feature spaces in geometry, which is all a bit abstract and something I can't really like bring my main to visualizing where there's a modifier. I like to see the data and insights and see plots and visualizations.

08:20 Yeah, I think having some kind of concrete project like that, just it completely changes how you experience it, right?

08:27 Yeah, definitely. And it's, you can then see, like actionable insights. And you can see the point where there's, I feel like going down a more theoretical math route is quite difficult to see. Or, for me anyway, like, what's actually going to be the point of this at the end, you know,

08:39 it's, it's a little bit how I felt as well, like, I felt like it was okay, now at some point, you just start solving these abstract problems for the sake of, well, they've solved them up to here. So the next logical problem someone's got to solve is to keep pushing it, but it's like, well, I'm not really sure what this is contributing back to the world, like, compared to so so many other things that you could do, for example, like healthcare or something. That was that was sort of my thought as well as lean. How about you?

09:05 So I when I, I did my degree in maths, I did pure maths. So that was even more abstract. But I quite enjoy it. I really enjoyed it. And it's only so after my degree, I didn't know what to do. So I wanted to get into accountancy. So at the time, I wasn't thinking about big data or anything like that. So I had the conversation with my family or project supervisor. So can I mention his name? He means a lot to me. Yes. James, Dr. James burridge. So we had a we had a conversation about what I want to do after I graduated, I told him about getting into an accountancy firm. And he said, Have you considered about you know, doing a PhD? So I thought in my head, I was thinking oh my god. And but I did say yes, I did say yes, because I really enjoyed I really enjoyed doing It green maps, especially my family project. I loved it. I really enjoyed it. So I did say yes. So in 2015, I went back to the same university. So the University of Portsmouth, and then I did my PhD on it. Now my PhD was in Applied Math. So that was aplicable. A big switch. Yeah, from the

10:21 No. shirt.

10:23 Yeah. And that's when I realized how important masses in a way and harm how many problems you can solve with mass. We, in my PhD, we actually came up with a model. So we derived a model from scratch, and we're able to implement the model in Python, and then get some data and and analyze the data. So we just became so real and so much interested. So yeah, I mean, that's when I fell in love with my data. And, yeah, even more math.

10:53 Yeah, absolutely. And the really interesting thing that I hear from both of you is like, there was this one professor who made such an impression. And I would say the same reason I went and spent a lot of time studying, what I did was, there was, I think it was like a calc three class or something like that. And it was just such a great person I connected with and encouraged me to go down that path. And, you know, I see in my daughters who are in their second year of going to be in their second year college, same thing, there was one teacher in high school that it was just such made such an impression. They're both doing their thing. And it's it's it's amazing how these small interactions and such huge effects. Yeah, yeah, definitely. Yeah. Jennifer, how are you?

11:35 The question is how I got into how did

11:37 you get into like, working with data and like, sort of computational programming side of things, just

11:42 the nature of the PhD? I did actually. So undergrad, never had been ever crossed my mind to do anything. particularly complicated. Just regular biology type. Statistics. p. Like t tests? Yeah, very.

11:57 Yeah, exactly. Yeah. Yeah. It just just enough to like, like, do the lab work. Right. Yeah, very few people, I think go into biology thinking I'm going to be a biology programmer type person. Right? Yeah. It somehow seems to like draw you in. Right? It like

12:11 totally accidentally, yeah. Yeah. Okay, when from t tests and ANOVA is on, you know, a couple of subjects to, to doing well, the my PhD was looking at appetite systems. And it was a PhD. That was the first is in Manchester as a first day. First time, they'd had two different departments collaborate on a PhD. So it was between a traditional body wet lab, which I was comfortable in, and the new Imaging Lab, so we use doing MRI studies, and like appetite systems, and of course, MRI, you've got three dimensional datasets. And if you take a brain image every three seconds, then you've got four dimensional data. And that requires very different and asked analytic approach. Yeah, that's

13:04 very different than taking measurements in a lab or growing something in a petri dish or something

13:09 very, very different. And if you go in, then you're in the world of signal processing, which wasn't what I was expecting. So now you've got to deal with drift of the signal over time. And then there's so many different ways to pre process signal signal data that what or time course data that not everyone agrees on how it should be analyzed. So that's fine as well. So yeah, the time there's several University developed software that's specifically designed for MRI. And so that makes it open source, this design is created with public money. There's three main software tools available, and then usually, like user interface, you know, use or click and, and build up your process that way. But they're also direct from the way that you can you can code it, instead, you can write a bash script that will then link all the different steps together. And then you can do batch style analysis. So I guess that was my first entry into that kind of analysis using bash scripts. And this was terrifying at the time.

14:11 Yeah, you're probably thinking, how did I get here? Like, this is not what I signed up for? I didn't what happened, huh?

14:18 Yeah. But it drew me in, like you say, so. Got into it.

14:21 Yeah, I think I mean, that's my theory about a lot of these things is, people don't go into the data science side of this in that, you know, the data science Python tools, necessarily, because they start out with the plan to be a data science programmer type. They just have their their doing work just like you described. And they're like, well, I've got to figure a little bit of this out, because otherwise, it's untenable, or it'll be, you know, 10 times more effective or whatever. And then a little bit more and a little bit more than you look back and like wait a minute, I've been programming all week. And that's all I've done. Like, how did this happen? How did it get here, right? Yeah, it's brilliant. This portion of talk Python to me is brought to you by data dog love Let me ask you a question. Do you have an app in production that's slower than you'd like? is its performance all over the place, sometimes fast, sometimes slow. Now, here's the important question. Do you know why? with data dog, you will, you can troubleshoot your app performance with data dogs end to end tracing, use the detailed flame graphs to identify bottlenecks and latency in that finicky app of yours. be the hero that got that app back on track at your company Get started today with a free trial at talk Python, FM slash data dog. I

15:32 think it's gonna be interesting, though. And like the next generation of data scientists, I think we've all come from a background where data science wasn't I think, a few years ago, like I will never have, it's a science when I was like, finishing off like my Ph. D. program in the last like a year or two, when I was looking for a career in Stanford see, like my peers a couple of years above me starting to apply for data science roles, where there's no, there's more data science courses. And I think I say we're going to then get a different breed of data scientists who have gone into wanting to be a data scientist, not necessarily come into it from a different background thinking all the way. I've been doing MRIs and data process and thinking process. And there are data scientists. I think that's so interesting. Yeah, go ahead.

16:12 Yeah, I totally agree. I mean, a few years ago, when I was doing my degree, I didn't know nothing about data science. I think it's it's it's a new, it's quite new. And it is sort of improving, there are new tools constantly being built. So it's it's spot. It's very, I think it's a it's an interesting area. So I think for anyone joining into the data world, they're realizing that there's so much to learn and to do. And it's just like, finding out what you can do the data. It's not. It's enjoyable.

16:45 Yeah, it's Yeah, it's really interesting. I don't think that for very long, there's been I don't know how many data science programs that are where they say, here's a Data Science University program. I don't even know if they've been around long enough for people to graduate from them yet.

17:00 I think master's programs at the moment, so one year, one year master's programs,

17:06 yeah, so if they're one year, people probably are, there's probably some graduates, but certainly 10 years ago, I don't think I think it was that way. So it's a really good point, Kaylee, that when you kind of look out and say, Well, what can I do you look at the degrees that you can go get, and then what those kind of lead on to? And you're right, that now I definitely think there's people that are like, Oh, this AI stuff is really interesting. Machine learning is interesting. The visualization is great. And when I was studying math, the choices for what you could officially do with your math degree were extremely limited. It was let's see, you could go into you could be what's the analysis for life insurance? Actually, actually, yeah, actually. Or you could go into like Wall Street. Or you could be a professor or a teacher. And like that, was it right. And I think that probably is happening across all these different areas. It's probably happened in biology. And it's happening in economics, and so on. And I feel like it probably this side of the world has opened up new paths for all those different focuses as well. What do you think? Yeah,

18:09 definitely, I think as well, it's quite as good though, because it means that like, we get quite a mix of backgrounds. So you do get the people that I know that sounds coming from a physics degree, biology degrees, I work at a company called peak where we've got a team of about 25, data scientists, and everyone's come from all these different backgrounds. And it's really great. Like the different expertise, some people have different ways of like, visualizing things, some people have come from more computer science taped group degrees. I think that's my point I made earlier, it's like, if you're then going to just get students that are going to go straight into data science, you sort of lose some of that background, and the expertise of different areas coming together. Yeah,

18:48 it sounds like everyone agrees on that. You're shaking your head, you actually don't want to see on the on the audio of it. Yeah.

18:54 Yeah, for sure.

18:55 And I think by working with people from different backgrounds, you get an opportunity to learn from them. So I think it's great.

19:02 Yeah, for sure. You touched on what you're doing now at peak with 25 different data scientists. Let's dig into that for all of you. So maybe you could expand on that just a little bit. Like what kind of projects are you solving what general airy working in and so on?

19:15 Yes, I'm a bit of a weird moments in my last two weeks of peak, so moving companies a couple of weeks times,

19:21 okay. Yeah, maybe tell us what you would have been doing and where you're going or what your future holds.

19:25 at peak. We've got 25 data scientists with a company in Manchester startup based company, who we sort of sell AI to businesses that don't have the capabilities in house to do data science and machine learning. We've got a wide spectrum of customers, ranging from like household retail brands like manufacturing. So within the data science team with 25 offers, we've got two teams. We've got one team that focuses more on like customer as a customer as things like recommendation systems, churn analysis, lifetime value, and then the team that is in is in The demand, as I said, we do things like demand forecasting, adventure optimization, warehouse optimization. So a lot of projects I've been working on are things like forecasting. So my background in time series analysis, so I did a PhD in change point detection, which I think is quite relevant at this moment, there are no other big change in life and like to be able to detect change points and fail to work with them. Just a plug there for change point detection. So yeah, so that's how I sort of come into this team, a recent project I've been working on for more than like warehouse optimization. So we've got a retailer that are trying to optimize how they pick their clothes and orders around the warehouse, and trying to like come up with like, the most efficient ways to walk around the warehouse to pick things up as quick as possible. So I've been working on it peak, I move into a company called resonantly and a couple of weeks time, who are another Manchester based startup, but they're more than they do like property technologies at prop tech. So there they've got like rentals. So a bit like I guess, Airbnb, but more so long term. rentals picture. Okay. Yeah.

21:07 Sounds like a bunch of fun projects. Jennifer, how about you,

21:10 I'm now working at at lad Bible Law Group. So they were born on a Facebook platform, now as a website, and on Facebook, and all the other social media you can imagine. Sure. So I'm the lead engineer down on that team on the day team. There's only three of us are tiny. And

21:29 current, you know that some of those small groups, though, are so cool, because you get to you don't get pigeonholed into a little tiny thing you get to actually see and, and work in the whole Yes, system, all the software. And it's, it's hugely valuable. It's a little stressful, because like, all of a sudden,

21:45 yeah, everyone wants something, which is great. It's a great problem to have. And yeah, I wanted to, I wanted to work in as a small company, so that I can be involved in all the things. I want to do all the things, everything interests me, I always want to learn new stuff. And this afforded me the opportunity to do that. So, yeah, there's just three of us. We're currently working on the new data warehouse using Google Cloud Platform. So we currently got our data warehouses that it's we want to have a GCP. It can be more scalable, and more robust. It's got great monitoring, and the dev team uses GCP as well. So it was good to be on the same on the same stack. Yeah. Nice. Yeah. Also, it's got you can use Python with it. So that's good. Yeah. So we've got big plans for that currently building out the data warehouse so that we can then carry out our big plans later.

22:32 Yeah, cool. The lad Bible. This is like, it looks like sort of a new take on news in the social media. landscape, something like that. Is that roughly right?

22:41 Yeah. Yes. I'm more lighthearted news, I suppose. Yeah. Spreading positive attitudes, which is really great. Like a real like working from the to teams are amazing. So great to work with. Yeah,

22:54 yeah. Sounds really fun. I definitely said that we could use as lean How about you.

22:59 I also work for a startup company in Manchester. It's called agent software limited. We aim to revolutionize the way estate agents across the UK market, their services using targeted prospective material. We've seen that agents have sort of seen sensational results with a fraction of marketing outlay. I work in a data team. So I work as a data scientist. I joined as a junior data engineer. I'm still learning about engineering, and I have a great manager and a great boss. He hates when I call Yes, but

23:40 he used to be my boss. Yeah.

23:44 Yeah. So yeah, we work on some interesting, great projects, we work. So for example, the project we worked on, about two weeks ago, we estimated the value of a property, so any properties in the UK, so we saw started model to get the estimated value of any property in the UK, which I think is great. And the one we working on the moment is the propensity to sell. And so that's the project we're working on at the moment. And I think at the moment, well, I've got some features that we both agree to use. So it is great. I'm enjoying working as a data scientist, especially within the company, because you know, there are some amazing people and I get along with them. So just makes makes life easier.

24:35 That's really great. It's interesting, I hear you all talking about all these people that you're working with how great it is, and people from who are coming in maybe, maybe they're in academics or something but they're not in the programming side. Maybe they think of the programming world as you know, you sit in a dark room by yourself and write code. And it doesn't sound like that's your experience.

24:56 That'd be great. I wouldn't

25:02 against that, there's a lot of people management, and not not like managing your team. But talking to stakeholders managing expectations. Yeah,

25:10 a lot of times you're the the person between the people who want the software, and and maybe the customers or who's using it. And you've kind of got to juggle all that, like that project management sort of side, right? Like, what are we going to focus on? What are we going to build for them? and so on?

25:25 Yeah, well, I mean, fit in our team. Anyway, we've got a project manager, and he's brilliant at helping to protect us from from that and making sure our priorities are aligned with the business. I sort code maybe half the time. And so that sort of planning and coordinating and meeting discussing discussions with people trying to figure out what they need, what we can do to make things better, and all that kind of thing. It's good.

25:48 Yeah, that's cool. How about you others, same time balance about or how's it, I

25:52 think at the moment, I did more coding. So I was a team lead until last summer. So I had like a lot of time just spent like managing a team and setting them tasks, making sure everyone like was okay with the customer work and sort of was that in between sort of firefighting type role. But then I took a decision that I actually wanted to spend more time learning and development. And that's where a lot of my Python development came up, because I was like, I just want to take a bit of time away from the team level and actually spend time coding, I think, from doing that, I've actually managed to carve quite a lot of my time, just focus on the coding side of things and actually developing the models and doing analytics. And then we have a customer success team that are sort of the link between our customers and data scientists. So they can do a lot more of the like, customer engagement and making sure that the customers are happy and sort of communicating what we're doing for them. And we get a bit more time to actually focus on doing data science, which is really cool. Yeah, that's

26:47 great. I

26:48 think at the moment, I'm doing more research, just for the more because we want to build a model for the propensity to sell. I'm doing more research, but at the same time we're doing research gives me sort of an opportunity to learn new new techniques and new tools. So yeah, I'm a bit I'm not coding at the moment, but I am missing it.

27:07 Yeah, well, one thing I think is worth talking about maybe is all of your backgrounds were not officially in computer science from the beginning your freshman year in university, and neither was mine. And that means when you get into the industry, it's it's a lot of like what you described. It's a lot of research, a lot of figuring out like, I know this, but but now they're asking me to do this other thing. I have no experience like what like I have no idea how to talk to a database. But apparently, this is what I'm figuring out soon. Right? That like that kind of half? Do you think that your academic backgrounds and just the whole sort of Advanced Study side of things just kept going when you got into the tech industry? And was it helpful? Jennifer?

27:49 Yeah, absolutely. I think, I don't know what people imagine on the whole, and they think about what's different between academia and industry. But there's a lot of transferable skills between PhD and industry. So planning experiments, or planning projects, that's the same research. So like, when you go into a new project, you've got to do the research to figure out like, what's been done before, what hasn't been done yet. So how could you go about testing something or learning something? That's the same, which, as you just described? And yeah, you come up with, you're giving you tasks like, well, I've not actually done that before. Like, I've not done recommendation systems before. Got to figure that out. You're still mentoring. So the higher up you go in academia, you start mentoring students, you do the same in industry, maybe

28:28 you're a TA teaching a class while you're doing your research or something. Yeah,

28:32 exactly. Yeah. And mentoring junior or foundation level data scientists, when you or whoever, when you get into industry, you're always learning, using your reasoning skills, really to get through any of the work that you do is the same in academia, and industry. There's a lot of overlap between the two, I think, definitely my self academic background has taught me to be sort of curious and keep asking questions. When I'm done tech, especially, it's quite easy. As a scientist, I think to sort of get your data, whether it's through some sort of like, psychic, learn type model, get the answer, and then just go with it. And I think it's, I think my PhD has sort of developed as that sort of deeper thinking of, but why is that the answer? can I explain that? What happens if I change something? And I think I've, like interviewed quite a lot of data scientists over the past couple years. And you know, it's more, sort of the ones that come more from a junior level that I've maybe just come out from last those haven't quite developed that sort of question. And because it's sort of what comes from doing a PhD in diving into much more deeper analysis on things that are sort of just to get the results. But then when you sort of dig into like, the question and say, why is that the answer? What How did the model perform? How does it work? The sort of as unsure about that, and I think that's something that I've sort of been trained to sort of keep questioning during my PhD but then back to tech is super ugly.

29:53 Yeah, I think it's the same for me. I think my PhD has really helped me to think things through and to question myself. And to check to double check and check again. And to test my work really, whenever I implemented in Python or even doing some analysis, I just it's I think it's, it's really helped me in that way. Because I think when I did my PhD, one thing I learned was, just because you have some you've done some mathematical analysis doesn't mean it's totally right. It's good to prove it by doing some programming as well, you know, it's much more, you know, it just makes it a lot more fun. And also, you're able to show what you've proved mathematically, you're able to show it as a graph, for example, you can, so people can actually visualize the math behind it, even if they don't quite understand the math. As soon as they see graphs there, they get a bit happier. Because, yeah, that's something I'm really grateful to I've learned whilst doing PhD. Yeah,

30:52 absolutely. It does really make it much more concrete, because math can be so abstract, and out there. And once you either make a program that can say do a recommendation, or can generate a picture, all of a sudden, it's like, oh, look, it's it's kind of real, it sort of exists, at least visually. Yeah. Yeah. So all right. So here's like, one of the big questions, I think of this conversation. You all talked about your, your background, and academics and studying and so on. And you're all working in the tech industry, not doing biology or statistics, or complex analysis, or whatever, you know, it is in the math world that you thought you might have been doing. So what was the thinking there? I could tell you, when I sort of dropped out of my Ph. D program, which I did halfway through, it was a big decision. And it like kind of weighed on me for a while it doesn't anymore, I'm all great with it, but a best decision ever. But you know, what was your thinking, you know, who wants to jump in first go for it,

31:50 I think I always sort of knew that I would probably end up in industry, particularly like, sort of halfway through my PhD. So because the way I can center is sort of set up, it was set up to work with industrial partners. So we had a lot of our PhD projects were with industrial funding. So I did mine with a company called the STL, which is the fence science and technology laboratory. And like within my PhD cohort, there was 10 people in my year, and everyone sort of had a different industrial partner. And within that they would industrial partners would come and presenters, we would have like regular problem solving days where they would come in in the morning. It's like a mini hackathon, where we'd get to work on practical problems.

32:37 That's so cool, I think really useful, right? Not just like you get here, and then you're on your own. That's great.

32:42 Yeah. So it was like a really applied pitch, even though we were spending time researching and dinner and work with Dr. Chen center, also on all these other events, and sort of things like conferences where there'd be half academic speakers and half industrial speakers, I think I was always more interested to go to the like industrial talks. And so people like using tech industry, because they're more applied, I could see the reason behind doing them. I think academia is he spent a long time sort of trying to develop novel it is and trying to prove them like asymptotically, or ever. Whereas I think I was much more to fire. I know the toes out. I'd like to research, what skill set to use, and then apply it. So I think it was always sort of more than applied to say, yeah, that's a big difference, right? You don't get credit for just necessarily finding interesting uses of existing technology. in academics. It's always about what new problem are you? question are you answering? Right? It's always a difference. Yeah. And I feel it's always very small, incremental changes. Yes, making someone's done it that like those like years and user research, and you might be changing a parameter or doing something quite small. I also think that academia is a bit sort of day to day of sensors, that slog so like, yeah, thank you publish, pay pose. And then it goes through like, Oh, your reviewers, and then you do your edits, and then eventually get your publication like submitted and published, that could have taken like two years, whereas an industry you could have your project in, sort of developed, like productionize thing ready and a few months, and then that value generated from it.

34:20 Yeah, it's a huge difference. Yeah, for sure. as

34:23 well. When I finished my PhD, I didn't Well, just before I finished, I didn't know what to do because I enjoyed working in academia. I was working whilst doing my PhD, I was also helping undergrad students with their math problems. So I enjoy talking to people explaining and just interacting with three students. I really enjoyed it. So when I finished my PhD I applied in academia and also applied in in the tech industry. So I was just said to myself, the first job that comes along is the one that will go for this So I first worked as a Python developer, I think there was really, I worked for a year. And that was, that was a great opportunity for me to really improve my Python skills. Because I was using Python on a daily basis. But I was missing my maths because it was only it was me. So I really missed it. So yeah, that's why I started applying for data scientists roles, and I didn't get one. But I think for me, I enjoy. I mean, I am in tech industry, and I really enjoy what I'm doing. I feel like I'm doing everything I've always wanted to know why to do because I do you enjoy programming, and I also enjoy math. So having both just makes me happy. At the same time, in terms of interaction, I do interact with my peers, so I'm not really missing out. But yeah, when I, before I finished my PhD, I didn't know what to do, mainly because I was wondering whether the level of interaction in an industry will be different to actually lecturing or Well, now that I do know, they're not not really different. So

36:04 yeah, one of the things I thought I would miss the most that I thought I was given up was that that teaching side of things, because I was teaching when as the grad in grad school, I was teaching all the levels of calculus and linear algebra and those types of things. And I really enjoyed that probably more than actually studying math directly. And I thought, Okay, well, I'm going to programming I'm definitely giving that up. And that seemed like a huge thing. And then it turned out that I ended up doing like professional development training stuff later on. And I kind of got it back anyway. And so it's not as big of a trade off, I guess. This portion of talk Python to me is brought to you by linode. Whether you're working on a personal project, or managing your enterprises infrastructure, linode has the pricing support and scale that you need to take your project to the next level, with 11 data centers worldwide, including their newest data center in Sydney, Australia, enterprise grade hardware, s3 compatible storage, and the next generation network linode delivers the performance that you expect at a price that you don't get started on the note today with a $20 credit and you get access to native SSD storage, a 40 gigabit network industry leading processors, their revamped Cloud Manager cloud not linode.com root access to your server along with their newest API and a Python COI just visit talk python.fm slash linode. When creating a new linode account, you'll automatically get $20 credit for your next project. Oh, and one last thing they're hiring go to lynda.com slash careers to find out more, let them know that we sent you. Jennifer, how did you It sounded like you were you were working a little more close to where your degree was, you know, when you said working in Baltimore, and so on, before you moved over to what you're doing now?

37:53 Yeah, so I did six years postdoc after PhD. And that was just more more MRI. It's brilliant. That's how I got into coding. And I liked I really liked zero still academia, but in, I guess the government research space. Yeah, still no industry. But yes, the narcissist has great experiments, brilliant, like, left all that stuff. But I'd always be attracted to the sparkling new technology, like this new technique. And I think I want to do that there's like, No, no, no, that doesn't answer this. You got to the question first, and then you pick the right approach. And that's still true, even in data science industry, but it's a little bit easier to handle now. Also, I'm aware of it now. So I know when to stop myself. But yeah, projects will be long. So we'll experiment from beginning to end of an end of potentially having finished the paper will be 18 months, really long experiments. So that I don't have a very long tension span. So that's not great for me. Also, once I've done the experiment, I've got the answer that I was looking for. But then you're going to write up the paper. But I'm interested in something else. Now you're

38:56 kind of done with it. But yeah, his work, you got to go through a peer review and resubmit all that.

39:02 Yeah. gotta write introductions and stuff and justify what I did. And, yeah, it's a really long process. And and yeah, I want to do the next thing by then. Also, there's not really too much feedback, you know, you submit the paper, and it's so niche, and then you get nothing back like what did what? Yeah, change anything? Did it progress anything, you might get some some references or citations later, but it would be, you know, a year or two, you might get five citations. And I would just did that mean anything? And there's some teachers might not even be positive about your paper. They might be saying this paper, this experiment was bad, and we're gonna do it differently. Yeah, yeah, exactly. So Exactly. It's very unhappy all around. So I didn't know going into industry would solve these problems at the time, but I feel like it's at least in the companies I've worked for so far. they've both been very small companies. The projects are much more shorter, the more defined they they worked in smaller increments. You get feedback quicker and directly with the people that you're doing. doing it for us. That's where you call and there's no papers to write.

40:03 Yeah, yeah. Do you think there's there's the opportunity to iterate. You've touched on this, that you iterate on something, you can go from idea to making a difference or making an impact in a month or two instead of, did you even make a difference? Yeah, the thing that really drove me crazy about the academic side was once you really get to the point where you're doing official academic stuff is so much so often, it's extremely limited. You know, there's 20 people in the world who potentially could care and half of them don't, you know, it's a pretty it can, it depends on the area, but it can be very, very small. I remember, and you can write software that is used by millions of people, or makes huge differences. And it's a big contrast.

40:45 Yeah, I agree. And also, like in industry, or like in programming, there's a lot more like open source. So you can get your function out there. And people can use your function, whereas it can be quite difficult to actually access papers. So even if you then published a paper in the journal, unless you also work in academia, you can't really access papers, unless you pay lots of money or find someone that can access it for you. Does that seem crazy to you that so much of this research, is government funded, publicly funded, but is locked up behind these extremely expensive journals?

41:15 Yes. drives me nuts. They get paid twice, because you've got to pay to submit your paper, and they got to pay to read it. Yeah. And it's peer reviewed by academics who are doing it for free.

41:23 Yes, exactly. Work on I know, some people that journal get paid and do work. But the reviewers that yeah, it's like part of their commitment to be an academic. Yeah,

41:32 yeah. I have feelings about that. Yeah.

41:39 That's another episode. Interesting. So that's really sad. You know, when I made the switch, it just seemed like a lot of what you were all saying really was part of it, I felt like, I wasn't going to make that big of an impact. I felt like what I was doing was actually much more interesting. I also thought that the employment story was really challenging in the academic space. For me, just talking to the people who are graduating a year or two ahead of me, how many applications they had, like I applied for 105 positions. I got five responses, two interviews and one postdoc offer for one year that I can reapply for like, Whoa, this is like a good university, have you got your PhD is there should be more than one offer entire realm of what you can apply for. So I'm just like, wait a minute, wait a minute. Wait, this is a little bit crazy. So yeah, there's definitely a lot of benefits. I one thing I do want to ask you all is, what would you What advice? Would you give back to people who are, say, in academics now? Not necessarily like, Oh, you should move on? Because whatever, because there's a lot of good stuff happening in academics as well. But what are some of like, the data science techniques or programming tips or career advice would you give to people who are maybe working in, you know, getting their PhD or master's degree or something like that,

43:01 I think I'd say try, if possible to get a little bit of experience in working in industry. So whether that's trying to find an internship or fellowship panel, some sort of PhD courses, or masters courses actually offer that where you can take like three months, maybe sabbatical from your PhD and go try doing something in industry, or those. I think in the UK, there's like ktps, which are knowledge transfer partnerships, I guess there might be some similar things over in the States, just to get that sort of experience of trying both academia and research and industry, but also to speak to people in industry, because I think there's like a sign to such a wide spectrum of a career, that there's different parts to data science, that may interest different people, and someone that's sort of spent a long time in academia and is possibly concerned and that few postdocs might find themselves wanting to go into more of a research post, an industry where the kids get off that high paid off stolen research for a bit by doing it in industry, which would then give them sort of like a foot in the door. And then they may Yeah, realize that either they want to stay sort of more research, or they might think actually the plates quite fun. Yeah, that's a good point. There's a lot of companies that have little research projects going on inside them. And you could definitely find yourself there as well. Yeah,

44:16 I mean, I would say just be open to new opportunities. So like, for example, if you're doing your PhD at a time, just make sure you take power into teaching, for example, just to see whether you'd be comfortable doing it as a long term, quick career. And also, again, like a Okay, and I said, you know, just go for at least just work in an industry maybe for one month or two months just to see if you're comfortable. And yeah, and also, just don't worry too much I would say about the money when you finish your PhD. Just make sure you get the right experience and just make sure you also learn new techniques and new tools. Because Yeah, I'm sure you get plenty of money.

45:02 Oh, yeah.

45:03 Yeah. Yeah, that's a great point. Because I think the hardest part about getting a job in industry data science programming side of things, the hardest part is the first job to go from zero to one. And once you've had that, that first job for a year or two, all of a sudden, many opportunities open up to reposition yourself or to get a big raise or move up or whatever. But getting that right experience is important. I agree. Yeah, Jennifer,

45:32 yeah, I think I almost don't have much to add to that is really good advice. But sometimes it can be, it can be hard to find people in industry to even talk to. So I definitely recommend going to meetups, you can meet all kinds of people who work for all different kinds of companies. I think a lot of my concerns about working in industry are ones that would be true in big companies, like really, really big companies that are not so that don't exist want to parent in smaller ones, for example, being siloed into smaller, very specific kinds of work. That might happen in a much bigger company, or stringency to software engineering processes. Or even just knowing that like some data scientist jobs are more r&d, which might be more similar to your academic style of working. And others might be more about identity engineering, or production using data science, but yet NASA's models, machine learning models, is that's very different kinds of working, that might, one of them might suit you better than the other. So just knowing that those are out there, and what benefits different kinds of work environments might have be good to know.

46:36 Yeah, there's a really big difference between taking a ml model and then writing an API to make it go versus researching what parameters you use to predict how the likelihood of selling

46:50 jam working on toy data. To get like the most accurate model is one thing, but then optimizing it to work for masses of streaming data is going on to some kind of car platform is a complete different problem, then you might have to make sacrifices of accuracy for speed or you know, any kind of there's, and they're all different kinds of problems. So yeah, you can meet or even I don't even want to go to industry, you might want to work for public service, or government, right? That's not necessarily industry, per se, but also in academia

47:18 does vary a lot out there, and you can meet all these people that meetups recommend, I think to add on to that is like meeting people like me up and trying to get an understanding of like, what the data teams are like, within companies, and especially if this is gonna be your first job out of academia, it's probably best that you find a company that already have a data science team off sites in the company, you don't really want to be the first data scientist because then you're not going to learn from anyone. And I think as well, like, if you've been in academia for years and years, and if you've gone like postdoc, maybe some fellowships, and then you decide you want to make that change, you sort of have to understand that you're no longer the expert, I think you build up quite a deep expertise in academia, that you've because all of a sudden become the expert in this sort of really, really niche field that only other like 20 other people are part of. But as soon as you go into industry, you realize actually, they were bored spec broad spectrum of data science skills, and you're no longer the expert. Even if you've got 20 years in academia, do you sort of have to take that step back and say, right, I need to learn this whole range of other skills.

48:24 Yeah, lets you identify where you really are interested in right. Yeah, yeah. Yeah. Well, that's, that's all really good advice. And I want a second idea of going to meetups and those kinds of events. Because the earlier you are earlier you are in that tradition, the harder and the more alone, it feels and like the less guidance I think you feel like you have and you can go meet other people who are in the same situation, or more importantly, maybe they're one year farther down the path, right, they've made that step and like, this is what worked for me. I also want to encourage people to offer to speak at these meetups, because I think that makes a huge, huge difference. I know when I started speaking of things like code camps and meetups and whatnot, it It gave me a much greater interaction with the community. I started getting job offers. I didn't know that I wanted but wait, wait a minute, really, I could go do this. It's not nearly as intimidating as it sounds coming from academics. You're used to giving presentations anyway. Right? So if you can find something focused enough meetups would love to have a 30 minute talk on something that you've been studying. And then now you're doing computation, right?

49:30 Yeah, I'm sure Kaley can agree that it's always having Kaley used to organize our ladies, right? Our ladies? Yeah. And ladies, man. Yeah. And I currently organized quite a monster. And yeah, looking for speakers. We would love for people to just offer us to talk on any of them. You have to be Yeah, always open.

49:47 Yeah. I don't speak for YouTube. But in general, I think if you were to say go to an organizer and say I would love to talk about this, but could you give me just a little bit of advice and guidance and feedback just to kind of mention For me of like, what is expected with this resume or just a little bit of help? for your first presentation? I'm sure you could get that help, right? Oh,

50:07 yeah, definitely. Yeah, actually, we were thinking about having a meet up event. I don't think we've done the details yet. But we were thinking of having a session just like that to train encourage people to give to talk. Because especially depending on where, sorry, yeah, especially depending on where they are in academia, and what kind of what they're studying, they might not have as many opportunities to give presentations. So they might not be as experienced. Sure.

50:31 Yeah, no, I think that's great.

50:32 I think in this audience, it can be very hot, like very easy to feel a bit like an imposter. Because there's so much going on that you think, I don't know enough to speak at these events. And what I found recently is actually finding what I didn't like. So during my PhD, I did change one section, I've done quite a lot of talks on that, because I found that when I do do talks, it's something that I know about, but it's also something that other people don't know about. So then they like to come and ask questions. So I think even though you might think your topic is only small or isn't as broad as some of the other data science topics, once you start to like present it and get more comfortable presenting something, you know, and then makes it easier to present on something you don't know as much about, yeah, you don't have time to learn about

51:15 it. It's easy to think everyone knows more than me. When I go on the internet, everyone seems so smart. I have no idea. But then in practice, you know, you go and you give one of these presentations. And you might ask like, Alright, who has a lot of experience doing this thing? I'm about to talk about your thinking everyone's gonna put their hand up, like two hands out of 20 go up, you're like, Oh, wait, these people don't have a lot of experience, I'm going to be able to teach them something. And I think that's actually really surprising and great.

51:42 Yeah, I actually think it works the other way. Because even if you presented something that there was an expert in the room, that's probably a good thing as well, because they can then give you the feedback that you might not have got from anyone else. And then actually like mentor you or train You're welcome into their team and say, we're actually doing something like this come and join us. Yeah,

52:02 but anyone else, but I find giving presentation helps me learn more about what I'm talking about as well, they I might, I might think I understand. But it's not to try to explain it that you realize there's holes in your understanding, and then you have to sort of fill in

52:15 those holes. Oh, absolutely. I think that getting some concept put together so you can present it clearly. You've got to figure out what is the essence of this thing. You've got to distill it down. And also you have to just have a different way of thinking about it. And I think the PhD research side of the world works really well here. Because if you're going to use some technology in a program, you just have to get some way to work. There might be three ways you could solve this problem. One of them works, you're done, right, like the problem solved. But if you're going to talk about is to say, well, there's these three ways and in this sewage situation, use this one, but that situation is that one and here's the trade offs and like that kind of deeper level research. I think it really matches well with those kinds of presentations.

52:56 I think one of my goals this year was to actually do you talk, but obviously Coronavirus.

53:05 Yeah. So that is so unfortunate. Yeah. So well, what were you going to talk on? Are we thinking,

53:10 Well, actually, I haven't decided, but I just want you to pick anything, if any, if I wasn't just to pick one topic and just do some research, learn about it, and then do a presentation and then talk to, for example, Jennifer, because she has more experience. It's done. She's done so many talks, meet ups and, you know, talk to someone who's more experienced and show her for example, my presentation and just yet to get some feedback. I love feedback. Yeah, yeah. And then see to go out there and give a talk. Yeah, so that was one of my goals. I don't know if it's gonna happen. But Fingers crossed. Let's hope all this just, yeah.

53:51 Me Up happening. So I'm talking about a virtual meet up on Thursday. So hardcopy, which was the event that this that inspired this podcast a few weeks ago because of Coronavirus, but they've moved it to a virtual meeting on Thursday. I don't know if that makes it better or worse, like speaking just screen where you can get feedback from anyone. But if they're going to do another investor, what you could like, if that interested, you can try and get on the speaker lineup for the one after Oh, thank you. Yeah, cheers.

54:18 Yeah, absolutely. I don't know if it makes it easier or better. I've done a lot of in person presentations, but also virtual ones. And they both have their own challenges, right. Like if you stand in front of a crowd of 500 people, that's intimidating. But also if you just stare at your screen and you have no feedback, you don't even know for sure if your your audio and video is cutting out and they're even hearing you like that's also really it could be distracted and hard. So yeah, it's it's different challenges, but I'm glad you're still getting to speak. That's great.

54:47 Yeah, I mean, today's Manchester started. We've had our first streamed live streamed panel event A couple of weeks ago, and we had everyone on a on a Hangout that there's three panelists and as much Seeing them the questions on a Hangout. And then that was streamed to YouTube Live. And I think that might have been it for the panelists as well, because they didn't have, as you said, they didn't have any feedback. They could only see the other panelists to see who they're directly talking to. They didn't. They didn't have YouTube open and have access to like, how many viewers apparently had or what comments are going down the side or whatnot. So yeah, it was quite odd.

55:23 Yeah. It's a challenge. You want to give a quick shout out to PI data. Manchester. You all seem like you're involved in that. And it's, if you're doing virtual stuff people could attend from all over even

55:33 Yes, let us know. As well, of course. And maybe you'll get two or more speaking opportunities as Yeah, we should be fabulous. Yeah, quite a lot. we doing? Yeah, we got monthly meetups and a monthly code night where you can bring a project and get help and get help. And we also do monthly podcasts. Super. Yeah.

55:53 Yeah, you have to give me the links to put in the show notes. People can check it out. Cool. All right. Well, ladies, we're getting close on time. So I think we'll have to leave it there. But it's really interesting to get a look at how you went through your academic side of things and then moved over and what you're doing now and that perspective, it's been really, really good. But before you get out of here, got to answer the two quick questions at the end. Which are if you're gonna do some data science, write some Python code. What editor do you use? Jennifer go first,

56:20 I jump around between three being on what I'm working on, or how I'm working atom spider and Jupyter notebooks. Or JupyterLab. Right on? Yeah,

56:29 yeah, sure. That's the next one, right? latest azulene

56:32 Yeah, I use atom spider and Jupiter as well.

56:35 Yeah, those are great. Kaylee.

56:37 I mostly use our Simon our studio person. Because I think Asher as the stuff I bought, when I'm tied to Python, I'm using VS Code at the moment.

56:46 Oh, yeah. Yeah,

56:48 I've heard good things about vs. Code, though.

56:50 Yeah, I really like it. I tried using. I'm not a huge fan of loopback. So everything seems to work. Well,

56:56 yeah. They've got some really good plugins. Yeah.

56:58 I don't think I've like fully, like experienced everything that can do. But I think I probably set up to run too much like our studio, so and probably not done things to pay

57:08 for anyway. There's so many options in that the all those extensions there. It's you can go. Everyone can have a different experience with the same thing. Pretty much. Yeah, for sure. All right. And then maybe one of you could throw out a notable package Python package that or a couple of you whoever wants to medical pie package that you've recently used that was awesome or worth telling people about?

57:31 I think Ilan is good. I think No, do you think

57:34 I say VEDA sentiment Vader sentiment analysis tool for specifically for Twitter, check their tweets. And so it's taking it's a it's an RP, but takes into account how people tweet, which is not like how people are fishing. Right?

57:50 Right, because you couldn't feed it like an article, trainer or an article and then feed it a tweet of like, well, this person's really frustrated because they are really terse. No, that's the limit.

57:59 If it's a word written in all caps, it takes that into account to wait the score, or how many exclamation points are used, or I think also emoticons or emojis. So it takes into it sort of understands tweet speak, as it will

58:13 be the sentiment Ah,

58:14 yeah, Vader Vader like Darth Vader, Vader sentiment of Vader. Yeah. The ADR sentiment project. Oh, yeah,

58:23 I see. Yeah, it's, it's on GitHub. I'll put it on. Put in the show notes. Very cool. Very cool.

58:29 Yeah, I was gonna mention Sublime Text. So that's more like a sophisticated text editor for code markup. And, and also data. Great. So data great is Yeah, provides context sensitive code completion. So it's good for SQL queries. And yeah,

58:47 so data grip is super cool. I mean, blows me away. Like you can do a query. And it just understands all the different structures. And then it's also built into pi charm. And in PI charm, if you write a string in a Python program, it'll autocomplete inside the string the stuff out of the database, which is less like that. Anyway. It's a good time to be programming the tools these days, and the libraries are so awesome. Yeah. All right. Well, that's it for our show. Ladies, thank you for being here. It's really been fun to talk to you. Thank you so much.

59:17 Yeah, okay. Yeah.

59:19 Yep, Dears, bye, bye. Cheers. This has been another episode of talk Python to me. Our guests on this episode were Jennifer Stark, Kaylee Haynes, and as lean become you, it's been brought to you by data dog and linode data dog gives you visibility into the whole system running your code, visit talk python.fm slash data dog and see what you've been missing. Throw in a free t shirt. Start your next Python project on the nodes state of the art cloud service, just visit talk python.fm slash linode li in Eau de, you'll automatically get a $20 credit when you create a new account. Want to level up your Python if you're just getting started, try my Python jumpstart by building today. apps course. Or if you're looking for something more advanced, check out our new async course the digs into all the different types of async programming you can do in Python. And of course, if you're interested in more than one of these, be sure to check out our everything bundle. It's like a subscription that never expires. Be sure to subscribe to the show, open your favorite pod catcher and search for Python. We should be right at the top. You can also find the iTunes feed at slash iTunes. The Google Play feed is slash play in the direct RSS feed at slash RSS on talk python.fm. This is your host, Michael Kennedy. Thanks so much for listening. I really appreciate it. Get out there and write some Python code