Software Developers Journey Podcast

#135 Emily Robinson making sense of data science for us


⚠ The following transcript was automatically generated.
❤ Help us out, Submit a pull-request to correct potential mistakes

Emily Robinson 0:00
What it means to be a data analyst at one company versus a data scientist at that company, you know, could be totally different in another company. So Lyft actually wrote a blog post a couple years ago now, about how all their data analysts became data scientists or data scientists became research scientists. Or at Google, I know I have some friends who are like, have the title data analyst, but are doing what is data scientists at other places. So there's also a lot of, I think, somewhat gatekeeping ideas around like, who deserves the title of data scientists like and like what you have to like, even if you have that title at company, people might be like, well, you're not a real data scientist. If you don't do machine learning in production, you're not a real data scientist if you don't use these tools, so on and so forth. Like I said, I became a data analyst ... funny story... actually, they are now called data scientists at Etsy.

Tim Bourguignon 1:04
Hello, and welcome to developer's journey, the podcast, bringing you the stories of successful software developers to help you on your upcoming journey. My name is Tim Bourguignon. And on this episode 135, I receive Emily Robinson. Emily is a senior data scientist at Warby Parker. Together with Jacqueline Nolis, she authored the book "Build a career in data science", which covers all the non technical skills and knowledge you need to get into and succeed in data science career. Emily, welcome to DevJourney.

Emily Robinson 1:37
Thank you so much for having me.

Tim Bourguignon 1:38
I'm very happy to have you, and somehow I'm stumbling already. So Emily, the show exists to help the listeners understand what your story looked like, and imagine how to shape their own future. So let's go back to your beginning, shall we? Where would you place the start of your tech or developer's journey?

Emily Robinson 1:59
I think for me, it really started in college. So it wasn't that I knew I wanted to become a data scientist at that point, because I don't think that was even really a job title yet. But I took an introduction to statistics class, in the spring of my freshman year with someone named Hadley Wickham, who is very well known in the art community, which is the main language I program in because he developed some of the most popular packages, including GG plot for plotting data, D plier, for analyzing it, so just, and really, this whole suite of packages, with contributions from now a lot of other great people as well called the tidy verse for doing data analysis, visualization, the whole pipeline and are so that I think would like really got me started. But I didn't have my first job in tech until, let's see, I guess would be five and a half years later, after a boarded trip to graduate school, where I was in a Ph. D. program decided academia was not for me, and took my masters and ran away to a boot camp and then into data science.

Tim Bourguignon 3:07
Ooh, there's some stuff to unpack there. So in college, you studied somehow statistics and already applied data science. Did I get that correctly?

Emily Robinson 3:16
Yes. So actually, I made my own major in Decision Sciences. So Rice University, where I went, I was it was a pretty rarely used option. But basically, I was in my sophomore year, I thought, maybe, oh, maybe I'll major in psychology, maybe statistics, but nothing really quite fit. And I realized, you know, I really want to do an interdisciplinary major that is at the intersection of psychology, economics, like the behavioral economics, field, statistics, and even some philosophy to really start understanding how and why do people make decisions? So that's what I ended up majoring in with a minor in statistics. And yeah, as you said, like obviously took a lot of statistics classes, which, again, I feel lucky I don't think this was the case a few years before I started college, but I was lucky that most of the statistics courses including included a programming component and are so I got to learn are along the way.

Tim Bourguignon 4:13
Is it something that is common in the US to just mix up your curriculum, like, like a buffet and take whatever you want?

Emily Robinson 4:22
I would not necessarily say common I know, at least a couple other schools have this maker on major option, but rice has about 1000 students, and each year like freshman, sophomore, junior senior, and of that maybe one a year makes their own major. So it's not common. So I had to write a petition, make a make a curriculum basically, which included some required classes, optional classes. I found three advisors from different departments to go over that curriculum justification provide input and then continue to be my advisors once the major was approved. So, no, I wouldn't say it's it's especially common, but it is an option that exists at some universities.

Tim Bourguignon 5:05
That is absolutely awesome. How did that go?

Emily Robinson 5:07
Yeah, I think it went, it went pretty well. So I was happy with that I ended up actually not being that far from the requirements of a statistics major, I think minus a capstone. So I guess I could have done that as like a backup plan. But yeah, I enjoyed it. I enjoyed working with my advisors. And then my senior year, I was thinking about what did I want to do after graduating I had done between my junior and senior year, a internship consulting internship where I got as you know, typical, I got an offer at the end of the summer, if I wanted to come back full time, after graduating, but I decided like, I don't know if I really want to do consulting. And so I was taking a close. Yeah, I think an organizational behavior class in my senior fall and an undergraduate version, but taught by one of the business school professors, and I was chatting with her I really liked the class, I liked the subject a lot, you know, again, organizational behavior is essentially psychology and sociology applied to work. So the psychology side was like, definitely in my wheelhouse. So I talked with the professor, I'd done some research, I've been a research assistant and some sociology and psychology labs, and I talked to the professor and she said, You know, I think you should consider going to graduate school. And she had gone to NC Ed, which is an international business school in France and Singapore, which only has MBAs and PhD students, although they actually they also recently started a one year Master's, but no undergraduates. And, you know, she said, I, she had graduated a couple years before. And she said, You know, I really enjoyed it, I think you should really think about academia. And so actually working with her, I decided, yeah, you know, what, let me let me try this. And so she wrote one of my recommendation letters, and I applied to NCS Looking back, I think, that was actually the only I knew I wanted to go abroad. I really wanted to live abroad. So and most and sad was, you know, in terms of looking for an English speaking school where I didn't have to have a Master's before. INSEAD was one of the is one of the top business schools outside of the US and actually including the US. And so, you know, that kind of rose to the top of my list, and I said, All right, let me take a shot. And I think by February of my senior year, I'd gotten in and accepted and start preparing to move to France and start a PhD.

Tim Bourguignon 7:34
How did that go? First time for a long time abroad, right?

Emily Robinson 7:39
Yes, yeah, so I've been lucky I had travelled internationally before with my family some for vacation, and actually had taken French in high school. Which I was so as I mentioned, in sad it's all English instruction because students come from all over the world no more than 10% of the student body is from one anyone country. So English is the language everyone has in common. That's what's what's taught and what's spoken among the students. But it was I looking back I'm like, very happy I did have some French because it made it easier to adjust to living and and France as you can imagine.

Tim Bourguignon 8:20
Being a French myself, Yes, I do.

Emily Robinson 8:23
Yeah, I had. So I remember talking with like, some so most students and those professors did not speak French. And I do remember some of them having some difficulties with like, landlords, especially if they lived in Paris, or just like getting someone to come and fix the plumbing I mean, your friends you there maybe this is maybe not limited to people who can speak French in terms of sometimes having issues getting things done. But I am very glad it was even so that the town instead is in Fontainebleau. You know, a lot of landlords their rent pretty much just to students, so they're, they're quite used to and the MBAs are only there for a year. So, you know, there used to like there was a fair amount of Apartments for Rent and English speaking landlords, but I still felt it was nice, even just going into shops and being able to say hello, and, and ask about, you know, my favorite bakeries like asked about the food in France, it was just kind of nice to feel a little bit more more like I was living abroad than just English.

Tim Bourguignon 9:22
Living the experience abroad. I guess. That's all part of the experience as well. I feel you. Feeling disconnected, and in the end having the feeling that somebody could knock on your shoulder and ask you something and you have no clue, whatever they said?

Emily Robinson 9:36
Yes. Yeah. So I was no, I really did help. Help me and especially when there were things there. I know, this is a this is a career development podcast. So I don't want to talk too much about France, even though I feel like we could go for a long time, but I will tell a very brief story of my Italian friend who like just that the train system is something else in France. So right we're like 45 minutes from Paris. So one time she'd like gone out to dinner in Paris and like bar was coming back to fund. It's just an announcement on the train. And I think because she's Italian, like she, you know, it's like somewhat somewhat of French, you could kind of understand it and it just ended up, like the train just didn't stop in our town and just kept going. And she ended up to this train station. It

was like 1:30
am. And she's like, asking the guy, what, what, like, what can we do? And he's like, yeah, there's not gonna be another train to like 7am. And she's like, No, you are going to like, call me a cab, fly me away, get me home, I am not just sitting in this train station for five hours in the middle of the night. So anyway, there was certainly some some very good adventures to be had. But yeah, so I spent two years in France a couple months in Singapore during that of the other campus. They we can talk about a little bit, but then as I mentioned, I kind of so PhD programs, generally you are the first two years, it depends on the on the program. But generally, the first years are a fair amount of classes. And you start in your second year doing some more research, you may have something called the comprehensive exams, like these often multi day exams, like essay based exams, or sometimes quantitative depending on your field, where you have to pass those if you want to continue in the program. And at that point, you got a Master's basically on the way. And I decided in my second year, I was like, you know, I don't really think I want to go into academia, which is what the vast majority of PhD programs are kind of set up for you to do. And I knew if I stayed past my second year, when I got my master's, it would be harder for me psychologically to quit. Because I kind of feel like oh, what what extra I spent in another two years here, and I got some knowledge, but I don't really have anything official, anything more official to show for it. So to me, it felt like that that point where I got my master's, was as much of a natural breaking point from the program as there could be. So I decided at that point. I told the program, I was like, Can I go on an indefinite leave that I may or may not still be on almost five years later? I think they've given up on me at this point. I didn't take the comprehensive exam. So like that, I think after that you have like, I don't know, 10 years to graduate who knows maybe seven. But in any case, yes, I, I decided to leave and move back to New York, where I'm from and do a data science bootcamp to try to transition into data science.

Tim Bourguignon 12:23
You said it took you five years to get into data science?

Emily Robinson 12:28
Well, now I'm including like my undergrad and the grad school. Yeah. So five years from my first stats, class programming. And our there's a whole sidebar here about like data analysts versus data scientist titles, I would still say I was in data science, but I got my first job. We can just say, working in data a couple months later at sea in December 2016. Okay,

Tim Bourguignon 12:47
So let's unpack that, again. You were as a master's degree graduate applying to a boot camp in data science, is that common? I kind of know, Boot Camps from a more web development perspective with really crash course on everything, web dev, but I guess it's the first time I hear from a boot camp in data science. How does that look like? And what kind of students apply there?

Emily Robinson 13:13
Yeah, so I'd actually say for the data science boot camps, having some experience past undergrad, whether it's masters, we had some people have PhDs or work experience, it's actually the more common thing, I think we maybe only had like one person out of my cohort of 20, who like came pretty much straight from undergrad. And so for me thinking about doing a boot camp basically was like art, I have like a good foundation in statistics in programming are, but I knew there were some areas I was missing. And so the key ones for me, then, are really four areas. So one was any kind of version control like Git and GitHub. So I wrote one of my first blog posts on one of my professors in graduate school, basically tried to use Dropbox as a version control system, and how that did not work. So well. I think I think things have maybe progressed since then. But academia also you can find people who are like languages that are pretty much only using academia or just like not using modern tools. So anyway, so I was like, Okay, I want to use getting good Ivan for and one of the reasons for that is, the second thing I was missing was a portfolio. So I had these academic credentials, but I didn't really if someone asked me like, Hey, can I look at your code, or you know, anything or like an app you made or an analysis you did, or visualizations, I didn't have anything to show. And in data science that can be something that is quite helpful is having some example projects when you know, that aren't these kind of super academic ones that are focused on publishing research papers that are a little more applied that can be really helpful to have so I was like, Okay, I want to want to do some of these types of projects, and then to publish those and share those on GitHub. Maybe through a blog, then three I had programmed in ours, I mentioned that the other big language, actually probably even bigger language use in data science is Python. And so I thought, alright, I would be good for me to learn some Python as well. And then finally, the fourth point is machine learning. So this is kind of funny, because some things in machine learnings are things that I had learned through my statistics classes, but by a different name. So it you know, there's I remember, I saw a talk on modeling once where someone highlighted like the Wikipedia entry for a technique that had the 20 other names that technique went by, like depending on your field. Yeah, so it's just yeah. Yeah, it's a big, it's a big problem, I would say in the field. But in any case, I did also learn some new things. Like I never done anything with text data. And so I learned natural language processing. But all that being said, I was like, Alright, I, you know, why do I want to do this, I want to learn these things. I think the network will be helpful for me. So from the alum who are working in data science, it's even more true now. And third, I just knew myself, I worked better in structured environment where there was a community, there were teachers that could help there was, you know, guidance on a curriculum versus say, taking three months and just, you know, trying to learn all these things myself.

Tim Bourguignon 16:17
Gotcha, gotcha. So no, no Kaggle experiment trying to solve problems and building your portfolio on your own and more structured and guided way to be sure you get through and, and have covered all your bases.

Emily Robinson 16:30
Yeah, what's the nice thing is the boot camp I did Metis kind of started out with like, we did five projects during it, the first one was a team one pretty structure that was just the first week, but from that on, it got kind of less and less structured. So the second was like, Alright, I can't like build a model to predict IMDb ratings. And the third one was, like, do some, like, find your own data and do like some kind of text analysis on it. And by our final project, it was totally open. So like, that was really nice, because it was a great way that people could showcase their interests or personalities. And then also, because people in the bootcamp had very different backgrounds. You know, someone, like we had someone with a PhD and lots of relevant experience was actually Part Time TA. And like her project was, like a pretty advanced one on, I think it was a lot of text analysis, it was like classifying wine, it was building an app, you know, because for her that was like her Challenge Level versus someone who say it was more, you know, new to the field. for them. What a challenging project is would look very different. So that was kind of nice. In a sense, it could be like, you could customize it for yourself, what you want it to specialize and what you want it to learn. But then of course, you're at the good the teachers and the other students to help refine it.

Tim Bourguignon 17:44
Yeah. And I guess you you get some cross-pollination as well, seeing what others do. And you get some experience from that as well.

Emily Robinson 17:51
Yeah, exactly. So it worked. It worked well, for me. I mean, I will put, you know, may what is the obvious caveat is that boot camps cost money. I was fortunate, so I moved back home, at that time, with my parents, and I was under their health insurance, I'd been making a siphon and grad school, but like, pretty much just sort of like a living stipend. So it wasn't like I was like, Oh, I was earning, you know, $100,000 at this marketing director, job, and now suddenly, I don't have an income, and I have children to support, you know, so on and so forth. So for me, um, you know, I was able to make that work. But, you know, I definitely do want to say, I know not everyone has the, you know, the finances to both pay for the bootcamp not work for three months. And then the reality being, you're probably also not going to be working for at least a month after you graduate, maybe more, because you only really can start applying to positions at the end. So I like Like I mentioned, I think after I was about three months after I graduated that I started my job at Etsy, so that was like six months with no income, which can be hard, is easier when you're a 24 year old living at home, who like does it has not, you know, have I guess lifestyle installations or, like, tilt up? You know, no one to support? Yeah, I just kind of like hung out at home kept busy with boot camp. Yeah. benefited from my my mother's cooking. That was definitely, yeah, that that's, you know, what made it doable, doable. And like I said, I recognize that's like a very privileged position to be in.

Tim Bourguignon 19:24
Good for you! Um, you came out after this boot camp? What What did you apply for? You mentioned just just briefly before I cut you, that analyst and data engineer, we can probably add data scientist and developer in the mix. The whole the four titles I see personally and in the cloud of data science, there must be some more you must know better. What did you qualify yourself or how did you did you call yourself and what did you start applying to?

Emily Robinson 19:57
Yeah, I think that's a really good question. Because, yeah, as you said, like data science is a pretty broad field, especially if you kind of include like, someone to do some, but also kind of falls under it things like data engineering. And the other interesting part is, in terms of like actual titles at companies, what it means to be a data analyst at one company for versus a data scientist, that company, you know, could be totally different another company. So Lyft actually wrote a blog post a couple years ago now about how all their data analysts became data scientists or data scientists became research scientists, or at Google, I know, I have some friends who are like, have the title data analyst, but are doing what is data scientists at other places. So there's also a lot of, I think, somewhat gatekeeping ideas around like, who deserves the title of data scientists like and like what you have to like, even if you have that title, a company feel might be like, well, you're not a real data scientists, if you don't do like, machine learning and production, you're not a real data scientist, if you don't use these tools, so on and so forth. So for me starting out, and I do think for even probably more true now is, like I said, I became a data analyst, funny story. So actually, they are now called data scientists at sea. There are basically two different data science teams, the one I was on, and then the one that is more heavy, like machine learning, a computer science or like PhD statistics level, kind of work, those are the people who are in those roles. But yeah, I do think that data scientists has shifted away, I don't know if it really ever was, but I think now it's shifted away from being an entry level title. So most people nowadays, it would be hard to get like a data scientist job right out of school, if that if that's not a PhD, maybe if you have a PhD, I think it's more common, you start as a data analyst, or there's even like, quantitative analysts, or maybe you're a data science associate or something like that. And then, you know, kind of transitioning into that data science role, whether it's by joining a different company, getting, you know, promote within your company, or just like, the magical you come in one day, and they're like, yep, you're all data scientists. Now.

Tim Bourguignon 22:13
Could you define a little bit what your definition is, of which and with what data sciences would do with that an analyst would do in your understanding of it?

Emily Robinson 22:22
Yeah, I think, for me, it's like, more help for me to define what I think are like the three types of data science. And so one of those is the analysis side. So that that's kind of the people who, you know, could have the data analysts could have the data scientist title, where you're focused on analyzing data for human consumption. So whether that's by building dashboards or creating reports, you are, you know, focused on getting the data in front of the right people. And then the second area is what I call decision scientists, which is a title some places, but I'd say be more, it's more common than just their title is data scientists, but the decision science work they do is making recommendations. So for example, if you know, your company leader comes to you and says, Where should we put our next factory? You know, that would be a case where you're not just like, it's not the type of question that's like, how much do we make last month? Right, we sort of produce a number you have to go and figure out alright, how, what kind of model can I build? How can I, you know, try to predict where what are the inputs that I need to understand where we should build our next factory. And then the final type of data scientists is called like, machine learning engineer, which again, could be your title, but your title might be data scientists call machine learning. And that's the the type of person who often will put machine learning models in production. So if you think of Amazon and how, when you you know, view an item, it has like a real where it says, like, you might also like these things, or when I was working at Etsy, if you search jewelry, while Etsy has more than 2 million jewelry items, how do we decide what to show you on the first page a result? That's machine learning model powering that based around? Essentially? Which, how can we predict what you will you know, that's you wants to make money, like any e commerce site? So how do we predict which, which of these items you will be most likely to buy? So, yeah, I think those are three areas. Some companies are pretty explicit. So Airbnb, I think is a great example where they often have a comma after the title, so be like data scientists, comma algorithms, data scientists, comma inference, which is sort of this decision science data scientists comment analytics, which I think is so helpful, because, you know, sometimes you do that by having different titles like data analysts, machine learning engineers, decision scientists, but sometimes you do that by putting them all under the data scientist umbrella, but I do think it's helpful to clarify wherever you can, okay, this is the type of data scientists that we're looking for.

Tim Bourguignon 24:58
That's make a lot of sense. way more than you looking at the titles. In the end, if I look at it from a system perspective, it's more kind of the, the machine learning engineer or side, being the working on the system making building the system, the decision scientists making sense of the system and the analytic side describing this interesting take on that. Oh, that's really cool. Thank you. I'm gonna research that. Very cool. So how did you get into Etsy?

Emily Robinson 25:32
Yeah, so, uh, again, talking about some sort of luck. So my brother Dave is in data science as well. And he, through him I knew someone, Heather, we Parker, who was no longer at Etsy, when I was looking for a job that used to be there. So I reached out to her. And she was able to introduce me to one of the old co workers who is a manager there, and he put me in the referral system. And I think this was very helpful, because maybe a little less than, but I think now, you know, there's just a lot of aspiring data scientists, human data analysts, you know, who don't have much work experience in the field. And, you know, some some I've seen companies are like hiring manager, I know, companies saying, like, we had to close a position, this like data analyst position after a week, because we got more than 1000 applications, right. So you can like trying to just get your foot in the door can be very challenging. And so a referral is not going to get you the job, but it is very helpful for like having a human at least look at your application. You know, again, it doesn't guarantee that you're going to get even an interview either. But in this case, they like looked at my application, they're like, Okay, this, you know, she is that's experience, she has these projects, GitHub, like let's at least have her interview, and do the initial phone screen with the hiring manager. And so I, you know, I did that and went through the interview process and ended up getting a job offer and yeah, and that's how I got started at sea.

Tim Bourguignon 27:07
Yes to networks. Definitely. And when I, when I tried to describe this, what you leave through this, getting your foot in the door, is really getting a chance to defend yourself, I have the feeling when when you're applying, you just you're playing mostly against a search algorithm or filter algorithm, if there's something in your resume that just don't click with the algorithm, that's all the filter algorithm, you're out. And you don't have a chance to, to to defend yourself and have a chance to explain anything, you're just out. And that's what networks help you do is just defend yourself at least on the first Echelon, and maybe you're out anyway. But at least you had a chance. And that's what's in this remote world. We're seeing now, more and more people applying to law to positions. And I really wonder how that's gonna work out? Yeah. Yeah,

Emily Robinson 27:57
I think that's a great point. Because right, there used to be some geographic constraints, right? So this job is only open to people in New York, which is a lot of people. But if you get remote positions that are open to say, anyone in the country, yeah, that's huge. And I do think that having a network when you're first starting out is tough. I think you can, you can work to build it through things. Like looking at your alumni network, from your university, maybe if there are people who went to, even if you didn't know them at the time that went to your same school and are now working in a company or in a job you're interested in reaching out to them, because I think people are more likely to respond to those who have something in common with. Or if you do a boot camp like boot camp alumni. But honestly, it is harder when you're not in the field yet. But once you are in the field once you can get that first job. One just getting your next job done, I think it's a lot easier, because companies look at that and are like, Okay, you've, you know, you've done something similar in another company. But it also puts you in a good possession position to start building that network. So for me, like going to meetups, and speaking have been away. I've met a lot of people and actually are, you know, again, how I sort of got my next two jobs. So the first one was, you know, this thing you can't really replicate, like, have a sibling who works in your field. But the next two were more on like, Oh, actually this person. For my second job. It was I had been on at data camp. I've been on the data camp had a podcast, and the podcast hosts, I'd spoken about running experiments at Etsy. And so he reached out to me said, Hey, would you be interested in doing an episode on experimentation at Etsy for this podcast? I said, Yes. I ran into him again, at a conference and he said, you know, our data camp is starting up. So we're going to start off like an experimentation team. I think you would be a great fit for the data scientist position. Are you interested and at the time, I wasn't really looking but he said, you know, please like at least talk about To the hiring manager. And I did and that's, you know, sort of then the process went from there. That's how I ended up at data camp. And yeah, I think your network, it's not just important for getting jobs, but it's also been, for me just so beneficial, like to have a community to have other people to, you know, sympathize with me, you know, when when, like, when selling is frustrating, or to, you know, learn from them about the newest art packages or to be inspired by their, their journey. So, it's, as you said, it's certainly harder now. With, with COVID and restrictions, but with the vaccines, you know, hopefully they'll start lifting by summer 2021. The Fall, and you know, we can go back to meeting in person, and I think it's just, you know, it's both beneficial to career, but also for me, it's been really fun. And just like a really, I'm so glad that I decided to kind of invest some time in building my community and data science.

Tim Bourguignon 30:57
Amen to that. I think exactly the same way. I would like to come back to one thing you said, or, you didn't say but I'm intrigued by this. You said that this person who was was hosting the podcast, got the info that you were running experiments at Etsy, did you work in the open? Did you make this public so that everyone's know that you're doing this? And that, then the first domino fall, that they know that you might be a good person to talk about it?

Emily Robinson 31:25
Yeah, so I did. So I started going to this data science meetup in New York, which used to be called New York hacker is called the New York open statistical program meetup. Gotcha. The organizer, Jared lander. And when I started seeing, he was like, Alright, you need to give a talk for our meetup. And I was like, Jared, give me like, a few months in the role. But after about six months, I was like, you know, what, I think I have a good talk to give because I'd, you know, had a fair amount of academic training and done like, research, like experiments and research, but doing experiments like and kind of psychology and organizational behavior usually had maybe a couple 100 people, you had to recruit them either in person or through services like Amazon's Mechanical Turk. was, you know, so I was like, Okay, I have some grounding, like some foundation ideas, like how many people do you need for an experiment, analyzing them, etc, etc. But when I got to Etsy, running a B tests, running online experiments, there were so many new challenges. It's like you, you have millions of people, it's working with stakeholders to define what we care about. It's having 1000s of metrics automatically computed for the experiment, which can lead to a lot of false positives. And I thought, you know, I really, I'd really like to share this because this, these are the things I I'm glad I'm learning, but I wish I knew six months ago, so I did my first talk, Summer of 2017, called AV testing in the wild, at this meetup. And, like, you're sort of implying that this was a great snowball effect. So I got an email a week later from the director of veterans mathematics director of a department at West Point who said, Hey, I saw your meetup talk. I'd love to have you come speak to our students and our faculty. So I gave it there. This also helped inspire me to submit a different talk for a conference. And yeah, like I like I said, this was how Hugo's the host of podcasts knew that I was doing experimentation at Etsy. And it's also how Julia silky who is now over at our studio, got to know her. So she talked with Lucas Vermeer, over at booking bookings, one of the like, biggest like, has sort of like the biggest companies like run a experimentation at scale. they've written a lot of papers on it. And so he was talking to her she was at Stack Overflow at the time about how they do experiments. But you know, I did it, she connected us, I got to know Lucas, and, you know, he recommended me for a kind of experimentation focus conference. So yeah, it's just, once you put yourself out there a little bit, whether through like a talk or a blog post, that can really kick off a lot more opportunities he doesn't need.

Tim Bourguignon 34:04
And I didn't want to put any words in your mouth, but that that's what I was wanting to do. I wanted to hear, it's not just like, it's, you have to make your own luck. You have to really get out there and give something from yourself if you're on the right subject at the right time, which is definitely a luck factor. But if you're doing that, this way, then it might take up and then and then something happens. And then you move on.

Emily Robinson 34:30
And there's there's really it's really hard to predict this. So, you know, one of the favorite things I've done, you know, my data science careers is writing the book you mentioned at the beginning. And my co author Jacqueline and I met at a conference where we were both speaking like random conference called day to day Texas. So Hilary Parker, you know, recommended me I had just spoken at our studio. She recommended me to the organizers, who because she was speaking as well. So I went and gave a talk. Jacqueline's in my talk. I was in hers. You know, sweet. We chatted a bit He chose to talk a lot. And then a couple months later, she reached out to me and she said, hey, I've been contacted by Manning about writing this book around data science careers, but I don't really want to do that by myself. Are you interested? And even though my talk at the conference was nothing about data science careers, I had written about it before. So I'd written about, like, some advice around like networking, like relevant blog posts with Jacqueline at scene. And yeah, I mean, that, you know, and so from there, I kind of wrote backtrack. And then I was like, yeah, you know, I'm busy right now. We talked in two months. And we did. And, you know, we started writing this book, having only met in person once. But it was a great collaboration. And now she's also really good friend. We're doing a podcast together. So yeah, like you said it. You know, it's not. It's not ignoring the role of like, locker privilege. But I think you know, what you said about, like, sometimes you do have to, like, make it or Yeah, open, open it up for those possibilities by putting yourself out there a little bit. And yeah, I never would have predicted on me to go to this conference, like find a co author for this book. I didn't. I didn't know I wanted to write but yeah, it turned out and and worked out, I think, really well.

Tim Bourguignon 36:10
Seems that it does or it did. As a father of three, I, I know, there's no good time to have kids. You never feel ready. And then at some point, you just decide that. Yeah, let's do it. Now, is it the same for a book? You never know that you're ready. You never know that now's a good time, and you just start writing and then suddenly you have a book in your hands.

Emily Robinson 36:33
Yeah, I think for me, I like like, looking back now. And I do think this is part of like pandemic fatigue. But I was like, how did I write that? was what I call it, there was like a 30 to 50 page book. I was like, wow, and you know, I and this was not part of my day job. I was working full time. So it's just like weekends, and you know, after work, but I even more amazed Jacqueline Rota, because she has a toddler. So that was, you know, more incredible. But yeah, I don't think I necessarily was like, Oh, yeah, totally 100% right to write a book. But I do think when it helped was two things. One, it was I had written some, like I mentioned some blog posts, that ended up. Like the topics I wrote about were like were addressed in the book. So things like building a network, actually, in the main for writing the book, I wrote a kind of post about, like, getting a first data science job. So I think it helped that I was like, Okay, I feel like I have something to say here. Because I get a fair amount of questions around this. I think, even though I hadn't been in the field a very long time, I did have that Organizational Behavior background where I studied things like negotiations and leadership, you know, doing doing well at work. So I found there was a lot I've learned there that, you know, things like the concept of sponsorship. So, you know, most people are familiar with mentorship and sponsorship, where someone is not just giving you advice, but they are offering you opportunities, whether by putting your name in for a speaking gig, or offering you financial opportunities, or if they're your manager, bringing up your name for a project that would be very high visibility in it in a big growth project. So yeah, so I, I felt like I had things to write about Plus, I honestly, like working with a co author was amazing, because Jacqueline had, so she did have a PhD, she'd been a hiring manager, she'd worked at a bunch of different with a bunch of different companies, because she'd been a consultant. So just had this like, really big depth of experience to draw on. And it also helped me personally honestly, like, keep me, I mentioned sort of liking structure, like keeping me accountable is, you know, working with her. And so we split up the chapter 16 chapters at each, and trying roughly for us to, you know, stay on the same pace in terms of writing chapters. So that was it made it more fun. It obviously reduced the work a little bit by having someone else be the writer of eight chapters, and I think made the book a lot better because we reviewed each other's chapters and we have somewhat different writing styles. I'm much more you know, oh my god, you know, we need to people need to know this, this and this, like, let me like, you know, lay it out, like try like as clearly as possible. Like here are things you need to know. Jacqueline's a little more like colloquial a little bit more kind of almost like story like and so I think that also we balanced each other really well by giving feedback you know, so Jacqueline might say to me like okay, like Emily there used to be like a little more of a flow here a little bit more story. This can just be like a bullet point list of facts. And me with Jacqueline sometimes saying, Hey, I think maybe this things that's five paragraphs would be clearer if we if we cut out this part, these parts and reduce it to three.

Tim Bourguignon 39:43
Amen for working in pairs and, and helping you through each other. Go out of our thinking processes. And I'm doing this with with one of my mentors a lot and just throwing an idea in and he just turns it 90 degrees and, and serves it back to me and say who I hadn't thought about this. And then let's twist it again. And let's twist again, after two or three hours, we landed two completely different position than where we started. And this is, most of the time, absolutely fascinating to see. And so I imagine that's can be the kind of work that you have with a co author away where you start building something, and then the other person expands on it and brings it in a new light that you hadn't considered. And which makes it even better. And then again, and again and again. But at some point, I guess you have to deliver the book, and, and stop this game. But when when did the book come out?

Emily Robinson 40:41
So the the final book came out, March 2020. Manning does what's called a meet zone Early Access Program. So I think the first five chapters released May 2019. And then like a couple more released every like four months after that as we wrote them. But yeah, so now it's been almost a year. I've done some talks, like kind of based on the book content, I mentioned, we had this podcast. So Jacqueline, I decided to make our own, you know, getting on the podcast game, make a companion podcast for the book. So one episode per chapter, basically, gonna serve two purposes. One is for people. So the book is not free. Although, as anyone who's written a book knows, we authors do not make a whole lot of the book sale proceeds. But it's, you know, through meetings, so for people who, you know, couldn't afford that, like, especially folks in some developing countries where, like the discount code, it's, it's, I think, $16 USD, but that can be a lot on some countries, or even for, you know, some aspiring data scientists here in the US. So anyway, so for people who couldn't afford it, or not sure they want to buy the book, the podcast kind of highlights some of the big points of chapters, but then also, it's a place for us to share some more personal stories. Because the book, we don't really talk about ourselves almost at all. And so the podcast is a chance for us to hopefully help some people by like, you know, adding a little bit extra color of Okay, what were what was this topic like for us? When did we face these issues in our career?

Tim Bourguignon 42:15
I'm not sure I've heard that before. Did you? Did you have an example of that before?

Emily Robinson 42:19
Oh, during this kind of podcast, actually, no. So I'd had this idea I tried to give I mean, there's certainly a fair amount of data science podcasts out there, but nothing I've seen with this format. But I guess what, like appealed to me about doing was, like I said, like, finding a way to like make the book, you know, the book content more accessible. Like we also have a lot of relevant blog posts. Yeah. And that I think I just started talking with Jacqueline about it. And first, she was pretty reluctant. But then she got really into it. And now she's the one that like edits all our episodes, which is great. And it's been really fun. Like, Jacqueline, I get along really? Well. I think we have a good kind of co host dynamic. Like I mentioned, we have different experiences, we have somewhat different personalities, but I think in a complimentary way, like Like I mentioned, so one thing that maybe exemplifies this is Jacqueline, the very smart idea of making short links for a book like rather than like manning.com slash build a crude, you know, data science, blah, blah, blah. And so we made two different ones. So the one I chose was data side. career.com writes fairly straightforward. And Jacqueline's, his best book cool. To the same page, but this gives you like a little bit of a peek into our somewhat different like, personalities. But Jacqueline is very cool. She actually won $100,000 on a game show called king of the nerds. like six years ago, maybe now so yeah, Jacqueline's. Awesome.

Tim Bourguignon 43:43
Okay. Cool. That's really cool.

Emily Robinson 43:48
Yeah, another podcast,

is http:

Tim Bourguignon 43:57
Do you have any advice for somebody starting to write a book? Since we are on books, and I have tried to write a book two times? So absolutely... for a friend. Yeah.

Emily Robinson 44:09
So as I mentioned, I mean, and this probably varies by person. But I do think having a co author, I'm just so happy. And I think if at all possible, like finding someone else. Again, for if you if you're someone who like that accountability is helpful, that's great. But even if that's not a consideration, just having someone to review your work, having someone else's perspective. Just Just that back and forth, you talked about was was great. For Yeah, and then besides that, I think, like people have very different processes. So some folks say for example, okay, if you're trying like you got to write like some every day, I actually didn't really write like that. I did more kind of Jacqueline actually did a blog post on because we wrote our book through GitHub, so she could look at the commits and the commit times about like how the book progress went, and I would say it was it was fairly steady in terms of like, month to month, but not on a daily basis. So like I often wrote, like a bunch on some weekends, sometimes weekday afternoons versus, and Jacqueline actually even more. So again, because she has a young kid, you know, she'd like go to a coffee shop and have these couple hours where her wife was taking care of their son until like, really focused. So I think that's good is to start thinking about Okay, how can I? What kind of environment do I know for something that is, you know, especially if you don't have a contract yet, right? So like, you don't really have external accountability, like figuring out hi can write that. And then I think the last thing I've heard for, like popular book authors, you know, maybe not writing a book unless you feel like you really need to like this thing that needs to be said. And I do think that's important. Like you, you really feel like this is I see these questions come up a lot. I give advice and one on ones, maybe I've read some blog posts, but there really isn't this comprehensive resource out there for what I think is very important topic. And I think, you know, relevant experience to, to share them to get that out there. And I think that, for me was like, the biggest motivator is just feeling like, I think there's a lot of people that could benefit from this book. So I think that's what really helped me write it much more than any, I think anyone writing a book is gonna find this out pretty quickly. But I would not do it for any kind of financial motivation, because, you know, probably a technical book, you're not gonna be on the New York Times bestseller list. And also, you could just in terms of money per hour, you would make so much more if you like, picked up consulting or something or I don't know honestly, even like a part time job as a barista would probably end up making you more money. So I, you know, it's it's nice, like, once you get it out, you got royalties, right? So it's like, eventually, income, you don't have to, like, work for so it's not nothing. But if, again, if you look at it, even like, probably for me, like five years out of writing the book, like okay, the hours I put in versus how much I've earned it, it was not a I did not write this for financial reasons. And Jacqueline was actually an independent consultant at time. So even Starker for her right. Like she knew what her hourly rate was like, she could go out and it could have been going out and like finding contracts. Like that is not why we wrote the book. We wrote it. Because Yeah, like I said, like just writing it for someone who like I really feel like this needs to be out there. Oh, and the last thing I'll say is, there's another book I'm really excited about by a will Larson coming out on staff engineering. So basically, there's been more books recently around people going to tech management, but I think less so for becoming a very senior, like staff level principal, individual contributor. And one thing he did, which we also did in our book was interview people. And I think that was just a great way. Yeah, obviously, it gets some additional content. But again, just if you're feeling like oh, I, you know, I know that. You know, I just really I feel like we're really missing this perspective from someone who's been in this industry or someone who's had like this path or this background. You know, doing for us for career focus books, doing interviews was a great way to to get those perspectives in there in the book and you know, not just have our two voices giving this advice, but like, we have more than 16 End of Chapter interviews plus like some sidebars throughout the book,

Tim Bourguignon 48:17
Thank you very much. This has been an avalanche of tips. And

Emily Robinson 48:24
That was that was a lot but I you know, sort of, you know, books, writing a book is really hard again, especially if you have financial responsibilities, especially if you have a job that's, you know, can extended to the evenings or weekends or it's just very taxing. So, you know, thing I always like to say is don't feel like you know, if you're if you're listening to this and maybe you're like senior level and you're like, oh, it just kind of feels like I should be doing a bunch of other stuff outside of work. Like that's why it's also okay, like not to it's okay to just be like, you know, doing and you have this podcast, of course you already do you know, something big outside of work. It's okay.

Tim Bourguignon 48:58
Not to forget the small pandemic we have on our hands.

Emily Robinson 49:01
Yeah. Oh, you know, laughing Yeah, yeah, like I think I mentioned, I've really, I've really cut back on speaking actually a lot just because, in some ways, because I don't have children, like I have more time in the evenings because there's nowhere to go. But I just have found I really do have kind of less emotional energy to just do extra stuff. And I think that's totally okay. And I just, you know, I've just decided like, you know, what this is going to be your career has seasons, and this season for me is like focusing on surviving the pandemic. I'm doing well my job doing this book podcasts. And that's mostly it.

Tim Bourguignon 49:37
Good for you. And that's the way it should be. Thank you very, very much. Where would be the best place to continue this discussion with you or start a discussion with you?

Emily Robinson 49:47
Yeah. So my website is hooked on data.org. That's not really like continuing the discussion. But if you are interested by some of the things I said you might want to take a look through my blog posts As I mentioned a fair amount of career ones. But the place I'm most active in terms of conversation is definitely Twitter. So I am Robinson underscore e s on Twitter, and tweet mostly around like data science stuff. And I'm also on LinkedIn, Emily, Emily Robinson, Warby Parker, you should be able to find me, I will say a quick note for that, which is probably a good general tip is, I do get a fair amount of connection requests from people I don't know that don't have messages, and I do not accept those. But if you, you know, you're listening to this podcast, and you're like, Oh, I want to connect, like, just write a short note thing. Okay. Or john, this podcast would love to connect or you know, if you have a specific question, and then I'll I'll happily accept and we can make 10 to the discussion there.

Tim Bourguignon 50:48
Awesome. Anything timely or not timely? on your plate beside the book? And beside the podcast? And?

Emily Robinson 50:55
Ah, no, I think the the podcast is really the biggest thing. Now, who knows? Maybe in a couple years, we'll I don't think someone asked if we'd write a sequel to the book. I don't really know if I have enough like, a different like enough different enough material to trade a sequel. But who knows, maybe in a couple years, we'll write v2 of the book, because there are some things that I don't think it's a critical act, but like I'm like, Oh, I think there's like, a couple things I'd like to add here. Or Oh, like one one thing is we have I think, a paragraph on Should I get a PhD, but I do not think was maybe strong enough. The podcast was just asked like, basically a therapy session for like, 20 minutes. But after discussing, unless you really, really, really, really, really want to be a professor, do not go get a PhD. It is a bad idea. For many reasons. But yeah, so right now the podcast is the big thing.

Tim Bourguignon 51:45
So you have the title of your next book, then.

Emily Robinson 51:47
Yeah, right. Yeah. Right, because I did read some of the like PhD, you know, PhD focused books. I had a 16 year PhD, but yes, my will just be it'll just be like a one page book. That's like, don't do it.

Tim Bourguignon 52:01
Awesome. Well, thank you very much. That was a blast.

Emily Robinson 52:06
Yeah, thank you so much. I really enjoyed it.

Tim Bourguignon 52:08
And this has been another episode of developer's journey, and with each other next week. I hope you have enjoyed Emily's story as much as I did. Her explanation of the 3 types of DataScience finally clicked in my mind. I won't have a puzzled look on my face when talking to DataScientists or DataAnalysts anymore, wondering what the heck they do for a living. And boy did we have a good laugh. Me cheeks still hurt. Tell me what you liked about Amiti's story, and what inspired you for your own journey. You can reach me on twitter, I'm at @ timothep, or use the comments section on our website under the transcript of this episode. Last but not least, give this podcast a rating on Apple Podcasts or wherever you get your podcasts. It really helps the show reach more listeners. And I am sure there is a data scientist somewhere analyzing those ratings, trying to find the ultimate answer to the great question of life, the universe and everything. Or is it a DataAnalyst? Oh boy I need to call Emily again...