The Care and Feeding of Data Scientists: Growing Careers
Hey everybody instead of the usual suspects I southern Bannon you just need. I am talking with Michelangelo Tax. Tina who's the the president of defiance shop runner which is a e commerce company in Chicago. mcglinchey thanks for joining. We talking about data signs management. This is something we've been China about for a while leading wrote and a Riley report together on this very topic a few months ago and and thought it would be worthwhile to get in and talking through. So that's what we're doing today talking about once. You have a data science team in place. How do you keep those learning? How do you keep them growing? How do you think about moving along in their careers? All that good stuff. So you're listening to the year diversions in our last couple of episodes which listen to strongly recommend that you go back and pick goes up by a little bit of recap here so we talked a little bit about getting into data science and the different types of data scientists. How you assemble a team and why it's so important to get the right blood in your team once you have? The team assembled or you. How kind of critical mass us most of your time? These days is not your managerial time by not beyond hiring. But it's instead on growing your not making sure that they're learning and they're sticking around. They're happy hair progressing in their careers. You say yes yes yes sorry you can't you can't you can't hear me now so and this was something that we it's your three chapters to talk about if I recall correctly because it's really important and it's a for me this the the stuff. That's not always obvious because sometimes if you're a good manager you're working super hard and but people don't see the things that you're you know fighting for the process of the you're putting in place two two happy I don't know maybe maybe it's a little bit cynical but sometimes given the only time you hear about is when people are unhappy with their happy sounds about right. Yeah I I think that's just as of nature of the beast so but when we were together we together for a couple of years. I remember even when I started at that job. There was a lot of. There's a lot of stuff that was in place when I write the action that you were part of starting some of this certainly part of of continuing the things that have been started before your time of that created a really nice learning environment and so I think that now upon on reflection I realized how important that was for me as a younger data scientist as a way of learning and and stay happy so if as you can get a little bit of detail about what was the you put in place out of things April so yeah it's funny. I actually don't know that I can take credit for having put in place a lot of the good things that we had But I can't take credit for stealing them and putting them in place in my current company actually But there's actually I think one of my favorite topics at least one of my favorite topics from the book and the way the way we talk about it in the book is kind of this idea of like foam. Oh like fear of missing out that like data scientists really have this feeling that like if they're not using cool new techniques learning things that they're like falling behind and just again anecdotally like I think a Lotta people quit jobs or go from company to company because they feel like they're getting stagnant or they're not learning anything anything or doing anything interesting at their company so some of these things that we had in place to help combat that that feeling that people can have so. What kinds of things are we talking about so we're big fans of of just a simple idea of having a journal Club where folks get together? Read a paper of blog posts. It's like something like that. Maybe they've ideally they've read the paper of the blog post but that's not necessarily mandatory and get together and discuss it Someone who has like selected the paper could maybe give the short presentation but really you can just do it over lunch and spend an hour talking about a paper blog post like learning something new dirty secret. You never read the paper. I know what I was GONNA say. Everyone has been very often but every once in a while I'll realized that Journal Club is on a topic that I wanted to do on podcast anyway and then I'll make sure gotta check. Oh all your listeners. Know the secret purchase but and we have like A. I think we find that a great way for people to keep learning stuff but it's also a good opportunity for the company because a lot of interested engineers or analytics people. Oh Com people who are a little bit Interested in getting into the field and it's just a good rates like socialized data science. Like around your your company One thing we've done a few times he's like when we forgot to select paper is to have like a cold. Like Davis is movie night where we will find a talk from an interesting conference. Like Pie de puts video video online a lot of conferences. But there'd be online and we'll just like sit for the hour and like watch talk a conference talk together and I think people often accumulate these lists of things they wanNA watch and they don't have the time for it and that's just another way to get people like watching something Learning something so journal Club thing. That were a big fan of the second one I think is like some sort of dedicated halftime and I definitely did not start this at our previous job and I definitely didn't even come up with the best iteration that I think that we've landed John so I've never three companies that have done halftime we've and seeing many flavors like one flavors like that we at a previous company was like half of every he Friday. You got to do whatever you wanted. Essentially another one has been like what is like a full day full day once a month. I think that's where we're doing a full day once a week at some point either way like that in those like short increments I find. They often don't work very well for two reasons. One people don't aren't able to at the time Apart like it was just half a day every week like you just get pulled in a meeting. So you'd get normal work like you can't actually. You can't actually do it even if it's like a day every other very hard at at the time aside so then we've tried whole team accolades and spending two days like working on something a team. We did cagle competition as a team wants. It's an and that has been fun has advantages but I think the the downside of that is that people can't work on the particular thing that they're interested or they want to learn about. Yeah Yeah So. That's a problem so I think both what we did at Civis and what we do now. Here is the idea of of individual hack weeks. And how often you do them depends is on your company. And what you what you can do. We do them quarterly. But you can. You can even do them once or twice a year but the idea is basically that each person gets to spend a dedicated week on like a hack project and because it's a whole week retire them to treat it like it's a vacation and plan for it like way out in advance and you know people can take a week advocation and get out of meetings and projects so do the same thing with this hack week actually set time aside a week as long enough to make meaningful progress on projects. It's like to actually get end to end on a prototype or try a new software package or open source somethin- and then the last bit is like some kind of bit of accountability to make sure people are actually doing something so Having a slight plan in place ahead of time that people can look at and then doing a presentation to the rest of the team at the end is super important so We do it on Friday as well help like a forty five minute presentation where the person that was on Hack Week. We'll walk the rest the team through what what they've done and so the rest of the team also gets to learn something new and cool but then that person realizes actually have to do something. Because I'm going to get off in embarrassments off. I have nothing to show for my week. When the Management I personally personally had basically I love hearing yet chancellor kind of live vicariously as interesting stuff? Yeah and and one thing. One thing we put in the report is that I think this is true of both both places. It's amazing how often something someone did is part of became part of like your normal tool set. I actually was a thing that got put into product or something like that. But it's worth mentioning that. Sometimes it takes a long time. So it's worth I think that's a good argument for happy. But it's also worth just kind of the expectation management. It might take six months or a year or something because it's just like the time has county. Yeah like you don't want to set the expectation that like tons of things are gonNA come out immediately and be useful but I think it is useful to like keep a list in the back of your head that was like hey someone introduced says to this idea and did hack week and then six months later it blossomed into this thing that we now use Because that's something you can point to for like justification for why you should do them poorly bay topic last topic in our report. There's probably pretty interesting voces heightening about how about career ladders and UH of formal career. Greta suggest well not just because it's important but like more informal learning stuff up so one of the things that we uh and besides in report was actually writing down what example career track. Looks like 'cause if it's something the thing that you've never seen before writing it the first time could be really hard and I think a decent amount of this stuff can be kind of teflonized so there are a few the things that I think are worth in this maybe to take them. In order so the career path we were on the other it in the lower levels Sort of junior in your levels starts out with one career track but then we breached at a certain point between vigilant trigger and management. I think we talked a lot about why we think that's importance. So maybe summarize summarize a little bit your thoughts on allowing people to advance in their careers without having to take on manage nurse Lonsdale a couple episodes back we talked about. I guess my transition into management which was very much like. Hey you're around and we need someone to be a manager like you see Mike. You should do this. Or they're not always not always an intentional decision for people to go into management and That's not always. It's not really a good thing. Generally like management is a very different job. And you don't want to force people to do it because it's neither good for the people that are being managed by that person who doesn't really want to do it and it's not good for that person because they don't really want to be doing it so like no one really wins in that in that situation so the message you we try to communicate as like hey you can succeed here and get promoted and have more influence across the company and make more money Cetera. Both by being like a really really good individual contributor and or by being a manager. You don't have to be a manager. In order to succeed at the company. He's worth steady backers for those who are in management and are wondering if it's if it's the right thing for them. Can you talk a little bit. About what love your job as manager largely is right now. It how it's different from just being a really really been a scientist. Yeah so I I'll talk about like the good parts of the job in maybe the not so great parts of the best parts of the job are being able to like look out across the organization and see opportunities for for problems for the team to work on to get resources for the team to work on those things but you really have to enjoy succeeding seating through other people like you're not like you don't get the adrenaline jolt anymore of like pushing some code and seen it have some effect or like thinking of an idea for an AB AB testing to see and there's some huge lift like you don't really doing that anymore. But like you might see two or three people on your team taken idea from beginning to end push it out and have great results. Then you have to be happy like vicariously through them. I saw like the best parts of the job coming up with ideas for projects getting them resource putting the right people people on projects and then like seeing them succeed. I guess but a lot of the time is like sitting and one on one conversations with people that you manage and like asking them about their week or asking them. I'm about their career progression or like writing performance reviews looking at spreadsheets of like you know how many people we have room to hire Like things that are very unglamorous. I watched a really good talk at one. Point that about management that said like what's both good and bad about management. Is that like the amount of things that funnel through you increases like you get to see more fun projects every like thing. Cool thing your team does have some exposure to and that's great but then on the flip side like every every problem or annoyance also getting funneled through and like if you have enough people someone is always unhappy about something or has some problem and it's like your job to kind of deal breath characteristic of Sun Wind. You make some really great individual a contributor. It should be like gutting to advance in their career through the ice each image. I think actually really just has to do with like. Do you really really like to write code or not. Can you be happy. Be Spending the vast majority of your days and weeks not writing code. I disagree with you a little bit. Okay so I think that I ending high-seas also naturally little rate less code as they get more senior but it's still regularly big me. I basically never wrote Code Tamer but I think I like a senior failure. Maybe that was a bad answer. But I think it's like an interesting one. It's a conversation like my tickets that I I see. A senior is still flipping across the Organization for problems that they are the primary owner of the technical solutions to the day. They're still close enough to the metal me technical decisions themselves rather than a a lot yet delegate trying to go in that direction because I think that the thing I didn't want to say is that you should go into management if you're good at dealing with people are working across teams or organizations musicians because I think senior individual contributors the really good ones they do need to be able to work across teams and they do need to be able to talk to other people and develop technical solutions and evangelized. I stuff so it's not really just am I good with people or not But I think that's actually. I totally agree interesting larger. The lesson here that I kind of learned firsthand would find myself having to parlay in in different ways to the people underneath that basically after point career progression doesn't come from being good at more technical stuff like that's important in the earlier stages I think taper off after a certain point and then being able to work with people because disproportionately or if we're even if you're in the ICU. Track I think that's right. Yeah I definitely agree with that but to get back to your other thing. I think maybe you are right that like even if more senior. Individual contributors aren't writing code all day. They are still thinking about deeply technical legal issues like they're thinking about architecture or like code best practices like things are still related to code and I guess on the manager side it is much more about like resource Out and maybe in project selection and some things like that so in your experience that once you go into management it never be an icy again. Or I don't I don't actually know the answer to that question. I'm or maybe they have a little bit afraid of the answer to that question because Do you remember do you remember. I said this I've said this a couple of times currently but I said at our previous job as well that if we ever required and I made millions of dollars that I would take your jobs again like I would go I would go and I would go and be like a member of the team again just like write code so I think in my heart. I like to think that you could go back to being an individual contributor and I do know people who have bounced back and forth back from management to being an icy again to go back into management both in engineering and in data science. I think it is possible But it is hard to keep your this goes up to date When you're managing it is I think it's possible to I think it but like you said the longer that you're longer that your hands are off-key order that you're far away from technical decisions? Yeah the heart rate gets too retro yourself back and legislate field so fast. Yeah but if you're a person who's been hired or thinking deeply about late hiring growth mindset folks and you know how burn anything instead you could probably get back into it. Maybe that's just what we tell ourselves so we can sleep at night so we have. These career paths that again. If you're really interested in the nitty gritty of career paths. You should check these out. We'll have a link to report ON DIGRESSION DOT COM. But I think the biggest thing that I wanted to emphasize kind of touched on a little bit Ra. which is the Netafim of these career? Paths that at dancing and data science after his point is not about whether you know more about the assumptions of ordinarily squares or how to write an API according to certain restful conventions or whatever. It's a little bit more about the the coalition building. Bring people along with you as you and I think knowing what kinds of problems to work on as you get more senior that here generally operating with more autonomy and independence. PAT's responsibility the other Meta point about the career ladder that's important it's just like the existence of a career ladder is very very important thing for team to have. I guess that like You can only get by with like vague Career advice in one on ones for so gore. People wanted to know like actually like what do I have to do like. Actually what does it mean to be a senior data scientist or data scientists or days manager like what are the the actual expectations and having a ladder at least gives you things that you can point to in those conversations and say like Hey I think you really granted these things things or maybe not so much these things. These are areas that you should work on. If you're interested in going this direction or that direction I think one of the challenges that I have with writing career. Career Ladders is knowing that there they become kind of this Sunday show document for all of these career conversations the choices that you have downstream so they have to be good and cover tasers and it's it's a contract of sorts between you and your team you're saying like if you do these things I'm telling you that you'll progress along a certain path so you have a big hole in it that somebody manages to walk right through kind of your fault as much as you may. You may remember this but I'm pretty sure sure you mostly wrote the career ladder at our previous job. Because for the reasons you just mentioned I was avoiding it for a really really long time and then eventually eventually there was a revolution from the people. And they're like we. You want a career they're called. We'd put a meeting on your calendar and we tried to give it a name that you that you say that you couldn't come so renamed something they deployments behavior but yeah I think because it has the seriousness where it is a little bit like a contract or like you've laid a stake in the ground It can be scary to right but it's also super important. Yeah but I think like the other other point I was going to make. Is that the way that you talked about careers. I think exactly the right one which is like it's the basis of a conversation between a manager and an employee it's in the beginning and end of everything that ever happens around curve. Russians League was that we read here and that we put a report are a little bit egg in part because we you know there's just a lot of people who might be taking these in adopting them for their own purposes but they're not very much more vague than some of the ones that you you were I might have written and I think that's because there's just so many different ways that scientists can show mastery and all of the different areas of expertise. They might have. It has to be custom crafted. Yeah I think I think the phrase that's in the report is like it should be more of a guidebook than an instruction manual like. It's not like a set of check boxes. Check check check check. I've checked every single one of these boxes now promote me like maybe in an ideal world. You can do that but it's always like a little bit more nuanced John's but like the letter has it. Gets you in the right direction. Like you have to be mostly in the right direction. Not Check every single thing rigorously general topic to sort of wrap this up and put on. There's a good phrase for this that we used to talk about a lot. which is the notion t shaped data scientists which neither you nor adding up but it's very very easily yeah? The first time I saw it was in an earlier O'Reilly report act ours called analyzing the analyzers that was about different kind of data scientist just archetypes but I think they have ripped that from this might actually be thing like in the human resources career literature or something but the idea. Is that if you imagine. Engine a capital letter t that So much it'd be brought across A set of skills and then deep in one particular area are Kerr. Letter is written where you're like. There are different aspects of the data science job that vary depending on what kind of data scientist you are but in my particular product engineering heavy culture. I believe we use 'em allen stats software engineering and then kind of a data systems are three different areas and as you progress S.. Like up the ladder. You're supposed to kind of have the the the the breath across the basics of each one of those areas which is like the crossbar tea and then depending on your particular secure emphasis like you should go deep in one of those areas and how kind of more expert level knowledge in like one of those specific areas. That says true what I've seen you from Washington number of of people that I've worked on my teams as they progress in their career like seeing that they actually start to become like a very very expert in the experiment design or they become really great engineers. Or they're interfacing with the product folks the and they understand that sort of stuff maybe. Does anything happens naturally but it's worth it's worth emphasizing like. Yeah there's a certain level of specialization. That kind of starts to characterize senior did being scientists. I think yeah. I think it's held like to ask you to ask the people in your team like hey which of these areas do you feel like the most passionate about you know. It's okay to not be passionate handed about every single one of these areas. Like I know some. You know I'm like I think I'm always the worst engineer on all of the teams that on like partly by design but also like that's just not the area that I am like most passionate passionate about other things. Well sure I mean it's expanding show exponentially quickly Eh. It would be impossible to maintain expertise in all of them anyway. But you still do want to have that common foundation that makes it possible to be a team. Exactly be able to talk to anybody somebody else. WHO's an expert in one of those places and generally hang with them? They're explaining things. Yeah well I think that brings us to the end of our time together. So the Hugh Michelangelo again for coming and talking was able to have a link sue the O'Reilly report on Linear Digression Dot Com even just download. All all of this. I think when they asked us to write it they said in for twenty to thirty pages ended up seven. Take Linear digressions is a a creative Commons. Never which means you can share or use it any way you like. Just tell them we said I find out more about this or any other episode of Linear digressions go to the near Digressions DOT com. If you like this podcast go and leave us review on itunes content. You can always get in touch with either of us us. Our emails are BEN AT LINEAR RUSSIANS DOT COM and Katie at LINEAR DIGGERS COM. In case you have comments or suggestions for feature shows you can tweet us at Lynn Digression. Thank you for joining us and we'll see you next.