Sean Taylor, Data Science Manager @Facebook, Lyft, talks about working in Data Science, benefits of working openly, shares recruiting tips
Sean Taylor has over 10 years experience as a Data Scientist, having worked as a Data Scientist and Data Science Manager at Facebook and Lyft. He is also prolific on Twitter (@seanjtaylor), regularly sharing his thoughts on the field, projects he's working on, tools he's using and problems he's trying to solve, and has gathered ~44K followers in the process. Sean has a PhD in Information Systems from NYU Stern School of Business.
In this discussion, Sean touches on a wide range of topics including:
How having a vast network of like minded people on Twitter has helped him
How working openly and sharing things that are still "Work in progress" has been a contributor to building this network and producing a better output
Types of projects he has worked on as a Data Scientst
How candidates can try to evaluate whether an opportunity is meaty enough
Ways to try and assess how empowered the function is when evaluating an opportunity, even though it's not easy.
Keep scrolling to listen to the discussion or read the transcript. If you can have any questions for Sean, you can DM him on Twitter or LinkedIn.
Detailed discussion transcript:
I have your LinkedIn profile open in front of me. And it looks like you've built a career for yourself in data science. And correct me if I'm wrong, but you've been in this space for close to 10 years now.
Depending on how you count, maybe even a little bit longer. I feel quite old.
You have been at Facebook, you were a research scientist, followed by being a research scientist, manager. You were there for about six years. And then more recently, you were a data science manager at Lyft.
That's a fair summary. I do also count some of the academic work as being data science adjacent to but those are the two main ones for sure.
You have 44,000 followers on Twitter, which is super impressive. Congratulations! How did that happen?
I’ve used Twitter for a long time. It has been a little confusing to me that I get so much attention in that environment. And I don't really make any special effort to grow that group of people. But I do like to talk about what I'm thinking about a lot. And I guess if I had to explain how it happened, it would just be that I work very openly, as much as possible. And I love telling people about what I'm doing. And I love asking questions. And I think maybe that just naturally simulates a lot of discussion in a place like Twitter.
Tell us a little bit more about this. Was it a conscious decision on your part to build up some sort of a following or maybe like your personal brand, as a data scientist, or it just kind of happened organically?
I never tried if that's what inorganic means? I do like to use it and I enjoy it. So it comes quite naturally. But there was never a point where I said, I need to grow the number of followers. And in fact, I think for a long time, I had very few followers. And I was using Twitter back in graduate school, it was just sort of talking about, installing Python packages and how annoying it might be or other more banal stuff back then. But I guess, using it for a long time, just in using it, I always tell people, if they want to have a lot of followers, you have to write a lot of tweets. So it just takes a long time to write that many. So having just used it for a long time, that's mainly where folks come from, but also maybe there's a little bit of a vacuum for hearing about what people are really doing in their jobs these days. And Twitter is a place where people can kind of find out, here's what people are doing at Facebook or Lyft. And, maybe there's something attractive about that, because it's hard to get that information elsewhere.
If you were to analyze your own Twitter stream, what are the kinds of things that you typically tend to tweet about?
One is definitely what I would frame as unsolicited advice, where you say here's something that people should do or something that is particularly successful. I have a lot of opinions like that, because I have worked for a long time, and I have developed some tastes. I do like to tell people, here's what people are doing wrong, or here's what people are doing right.
I also like to use it to ask questions and almost treat it kind of like a search engine, where the responses are generated by all the people that I know are really smart, which is a really great tool, and it's one of the luxuries of having a lot of followers is when you ask a question, it's pretty easy to get a lot of responses. And that can be really valuable for drumming up ideas or trying to find examples for things. Just a couple of days ago, I asked about things that people had invested in learning that they felt like didn't pay off. And it really stimulated a lot of great responses. I learned a lot from reading them and it was just kind of fun to be able to do that. That's another thing I use it for.
And then just staying in touch with people and finding out what they've up to has been probably one of the primary values. Knowing what people are working on and what they're interested in, just helps you stay on top of what things are happening in the world outside of your little bubble. And these little streams of information have been really valuable in my career. It comes up a lot that people ask me, how did you find out about that paper or that package or that idea? I found out about it from someone I know on Twitter.
Has your personal brand as a data scientist helped you in any way?
I'm very interested in causal inference. And it's hard to say if the personal brand has caused my life or my career to be better in any way, because I don't know what would have happened if I didn't have a Twitter account. I can say that I haven't changed jobs very often. After graduate school, I worked at Facebook, and I didn't have a large Twitter following back then. And then I got my job at Lyft. Maybe the person who hired me knew about me from Twitter, I actually don't know. But I can't I can't say that I found a new job, or someone has found me through Twitter. So career wise, it's hard to say that it's been a big deal for me.
I think that awareness of my work has probably been one of the biggest benefits. So for instance, I do write papers and blog posts. I have a Python package that I worked on, and an R package. And I think having a Twitter following and a personal brand has been useful in getting people to pay attention to those things that I work on. And you know that's certainly valuable to have other people know about what you're doing and either to give feedback or to use it themselves and grow the projects, and kind of help you do better on your next project. So, the personal brand, mostly I like it because it helps me connect with other people and find them and they find me and those things end up just kind of creating better intellectual growth for me that would be I think the primary benefit.
I was also pretty early in meeting people in real life that I knew through Twitter. I was doing that even very early in my Twitter days, because I lived in New York City, and I was a graduate student, I didn't know that many people. And it was easy to meet people that I met on Twitter. And, it's turned into a really big network of people that I know. I can't say it’s like the source of my career success or whatever. But, it's been really enriching for me to just have a community of people that I don't think I would have met them in any other way.
One of the things that a lot of people who are interested in this field do potentially grapple with is that, maybe I do not have a traditional background you might need to become a data scientist. But I am really interested in this field. And that transition can be a bit challenging - how do you demonstrate you have the credentials?
And in some ways, what you have done - you're doing all of these projects, and you're sharing your knowledge, and you're doing these things on your website on Twitter, is amazing proof that, clearly this person is interested in the field as well as knows what he talks about.
Sometimes I feel about this, to use 44,000 followers to tell them about it, there's, some impostor syndrome that creeps in. But I do think that working, the theme of me, using Twitter is a little bit of like, kind of working openly, and talking about what I'm doing and thinking about, which is also something I do within the companies that I've worked in. And I’ve found that to be a really valuable tool, some people try to take their work and, keep it within their own brain and their own computer with a couple other people. And I find that for the kind of work we do that just being in discussions about it constantly. And being really curious about how other people would think about it has been a really valuable tool. And it's really hard to explain it. But I just think that there's this openness to research work. And when you're really intellectually engaged by something, you almost want to be able to talk to people about it constantly. Because sometimes people call it a rubber ducking, where you have somebody that you're explaining something to as you're thinking about it, and that helps you clarify your thoughts. Maybe the discussion is useful, just because it forces you to articulate how you're thinking about something so other people can consume it. But I just find it to be a really indispensable tool for working on problems that I've faced.
That's a really interesting point. Can you share examples, either from projects that you’ve worked on inside a company, or just through having this following, where these discussions have led to very different, or significant changes in how you were thinking about a certain thing?
There's one that comes to mind immediately. I worked at Lyft, most recently, which is a great company. And I really enjoyed all the interesting problems there. It's just a company with so many interesting research problems that I felt like a kid in a candy store. And one of the ones that came up very early on was how do you represent spatial data within the machine learning system. So it's a little bit of inside baseball, but roughly when you have latitudes and longitudes and you want to put them into all your machine learning models, that's not a very good way to represent them. And so there's all these ideas about how to represent spatial data as features in a machine learning model. And I thought about a lot of different ways of doing that. And I just started writing documents internally at Lyft, to clarify that how we were currently doing it, which was using this technique called “geo hashes” felt like it wasn't the right idea, and there were some reasons why that was kind of causing some problems in the models that we were building. And I wrote this note internally at Lyft, called “Death to Geo hashes”, which was me saying, this way that we're doing things isn't good. And we need to think of a better one.
And that stimulated a lot of great discussion from other people. And someone had one idea I was really enamored with, about how to do that and then eventually actually got on my team. Just as a result of all of us having discussed that someone came up with his own idea about how to do it, which is using a technique called Gaussian mixture models. And it turned out to be a really great idea. And it's something that I never would have thought of, and something I actually hadn't even heard of before he proposed it. So, starting a conversation about something that you're working on, it's like an improv idea of people can riff on it, and then that dialogue is what really creates the value, because a lot of the best ideas aren't going to come from your own head.
And do you think that some of it, what you're saying is also about being okay with sharing ideas that are not necessarily fully fleshed out? People sometimes struggle with feeling vulnerable in terms of sharing something which they are not so confident about yet. And kind of being okay with that.
I couldn't agree more that there is a vulnerability to talking about things before they're done. And it can be really limiting to kind of hide things until you're done. And it's something that you have to get over. It's actually a big theme for me. I've been a manager at two companies now, and one of the things I coach people on a lot is sharing their work more openly as they're doing it. And one of the ways I frame it is bringing people along for the ride so that when you do work openly, people are more likely to believe that you did good work, because they saw the process by which it was made, they had some say in it, maybe early on, and they could help kind of provide early feedback, and they just feel like they're part of it, because they got to experience some of how it was done. So there's a lot of value created. But it does require this showing people stuff that you're not quite ready to show them yet or admitting that you don't know the answer to a question you've been thinking about for a long time. And that can be tough to do. And I think particularly around really smart people who are intimidating.
I gave a talk last year at this causal inference workshop, which I felt very flattered to be invited to because the other speakers at the event were really famous academics that I really always looked up to. And they invited me to come and talk about some of my work at Lyft. And I talked about a problem I hadn't solved yet. I had this big worry that I was going to show them this problem I was working on and they were gonna know the answer. We already know the answer to that, or the answer is something that we could have solved really quickly. But instead it was the opposite. It was, your problem was really interesting. And, we'd love to talk to you more about it, which is exactly the response I was hoping for. But it did take me being a little vulnerable and being able to share something that wasn't really done yet.
I really like that. And do you think that your openness or your willingness to do this is perhaps one of the key ingredients in terms of why you've been able to build this following? It's almost as if - since you're asking questions and sharing these things that are not yet complete, you're inviting participation, and solutions from your audience?
I certainly hope that that's the impression that people get. I will speak to my experience. And I'm willing to do that. But I don't think I'm really ever willing to say I know the right answer about anything. Because I know a lot of smart people that know a lot more on any topic I can think. So there's this humbleness that I think has to come from hanging out with people that are really good at stuff that you end up just thinking there's no way I know the answer to this better than other people. And you have to just kind of come into things with a beginner's mindset, almost no matter how much you know about them. And that's kind of like in the spirit of I was trained as a scientist. I think as a good scientist, you're always questioning whether you're really right about something, even stuff that you have studied for a long time, you have to be open to the idea that the way that you think about a problem could be over overwritten or overturned in some way when there's new evidence. And that's something that I really internalized into the way I think about solving problems and studying things.
So, let's say someone looks at you and says, this is so amazing. I would also like to have, not necessarily a following, but sort of this large group of people that I can bounce ideas off of, or pose questions to and discuss things that I'm working on. How would you suggest they go about that?
Probably the easiest way is to work on finding a small group. To get a large group of people, you probably have to start with a small group. So, when I was in graduate school, I had a couple of people in my department, my lab mates. So I had a small group of people that I was studying the same things with, at the same time. And that was a valuable early community for me. So the people that are around you that you're working with, they're probably your best candidate, because they're thinking about the same things. And those are the people that you want to be having a dialogue with. And I know people do create these kinds of small groups for, but often they're kind of transient, it might be just for the course that you're doing or for the project that you're on or something like that.
So maybe the real part of it is how do you turn a small transient group a little bit longer term that you can carry with you for longer, and that requires being a little bit socially proactive, you have to be willing to organize meetings with people or chats or discussions, and I'm very extroverted. So it's easy for me to do video calls with friends, or people that I've worked with in the past, or to reach out to people and try to kind of coordinate some time to chat with them. But you do have to put energy into community building and to have people to discuss things with, because it doesn't come just for free. It's something that you have to invest in.
Do you think Twitter has been a good medium for what you have to say?
That's a great question. I don't know, if there's anything intrinsic to Twitter. Certainly, the need for brevity is a useful one. And having to compress the things that you're thinking about into short space, and that's kind of inherent in the medium, is an interesting constraint to operate under, and it forces you to clarify what you're thinking about really carefully before you write it.
Also, you're a little freer with sending things out, because there's a little bit less of a burden of having to craft something perfect, and there's a little bit of imperfection that is inherent in the medium. And we all have typos in our tweets and that we can't correct. And we just kind of have to live with that.
So, maybe Twitter does optimize a little bit for just kind of creating opportunities for people to spit things out and riff on each other's stuff a little bit. And there's also this, just it's a place to kind of just hang out while you're bored and waiting for some of your machine learning code to continue to run. So there's a little bit of that there.
But lately I feel like I wish I had invested a little bit more in a long form medium. And I probably would like to start blogging again, a little more actively, because I do think that there certainly are some limitations with just having to write everything in short form that we can't have the kinds of conversations I'd like to have some time. So that's something that I think is worth identifying as a limitation of the medium, we just can't have certain kinds of conversations there. But Twitter is a great platform. And I really love it a lot. And it's been a pretty big force for positive stuff in my life.
Then, let's just quickly switch gears to a career as a data scientist. You've touched on this a little bit in terms of the kinds of projects you've worked on. But if you were to describe the role of a data scientist, to an outsider, how would you describe it?
We could spend the entire podcast just on that question, because there's a big diversity of roles that are all called data scientists. And that's one of the primary challenges that we deal with, as within that role is how do we define and scope the work that we do to contribute toward the success of a company or any kind of organization?
One way that I think about it is that data scientists produce decisions. And, that's a very, you have to think about decisions very abstractly, but decisions are things even when a page is rendered on Google and the search results there are based on all the decisions that Google has made about what to display in the search algorithm. Or should we launch this version of the app? Or which market should we expand into, or all kinds of decisions that companies face on a regular basis, and they need some principles and procedures for deciding what to do. And that's mainly the way I think about the role of a data scientist.
And it turns out that the best way to make decisions is to bring some evidence to bear on decisions and evidence that we have is in the form of data that either happens internally to a company or that we're able to kind of bring in from the outside and then synthesize that evidence and estimate the things that we need to estimate and then ultimately landing on what should we do as a result of the evidence that we have is primarily the way that I think about the role of the data scientists.
Someone who's not familiar with tech might struggle with the distinction between the role of a Data Scientist and an Analyst. Is there a distinction? And if yes, what is that distinction?
The distinction between a scientist and an analyst - I'm not sure I'm willing to make any claims about that. I think sometimes those titles are used to kind of create artificial distinctions between people that are really working on very similar things.
Maybe there's some difference in qualifications for the role, maybe, somebody has a master's degree or a PhD and somebody else doesn't. Or maybe there are distinctions that are made about what tools people use, or what level of the stack they're operating in. So there are some people that use like Excel to perform the same task that someone else performs in SQL that someone else performs in Python, or R and but they're all fundamentally doing the same things, or they could be doing different things, and they just have different ways of approaching the same problems.
And then, the other problem with the data scientists versus analysts distinction, or even all the other - data engineering would be another one I would throw out, or that they operate at different levels of the stack. So data engineers are concerned with making things run reliably, and just making sure that clean data is available for other people to do things. But that's completely necessary for people at higher levels of the stack to do their jobs. And then there are people who are data scientists that are even building apps and websites, and that people can use to kind of access and make better decisions. So there are so many different ways that people work and qualifications that they have but ultimately, we're all trying to do the same thing, which is take the data that we have available and make a positive impact on the companies and organizations that we have.
One of the distinctions that we came up with at Lyft that I like is whether the decisions that you're trying to influence and improve with what you're doing - are you changing a human decision or changing an algorithmic decision. And that was one that is pretty valuable to think about. If you're a data scientist that's improving algorithmic decisions, that means you have to change the code that runs, that powers your apps, or websites or production processes in some way. And that requires having software engineering skills and interfacing with software engineers. And then there are people who make improved decisions by interacting with humans in some way, and maybe just convincing them through a chart or through a report that they write or even just through a conversation, that one course of action is better than the other. And that's fundamentally kind of a different landing point for your work and changes the way that you have to work in a pretty fundamental way. So that's probably the most useful distinction that I would come up with. But the titles are probably the least useful way to think of the distinction.
Can you share maybe an example of an exciting project that you've worked on as a data scientist in your career so far?
There's been a bunch of them. Maybe one of the most ambitious ones that I worked on was at Lyft which was called decision science products. And when I first arrived at Lyft, there was a weekly planning process at Lyft, where we had to decide how much money to spend to balance the market. So Lyft has this problem that there's a supply side, which is drivers, and a demand side, which is riders. And if we have way more drivers on the road, then we have riders, then they all sit around, and they don't make any money. But the riders have very fast pickup times because there are drivers everywhere. So one side gets benefits but the other side doesn't.
But if you have a lot of riders and no drivers, then all the riders are waiting and not getting rides. And the drivers are making tons of money. So our goal is to manage that market and kind of balance it. So have the right number of drivers for the amount of riders that we're expecting. And the way that Lyft manages that is by spending money to simulate different sizes of the markets, there's almost a little Federal Reserve Bank internally within the company. So we can send coupons to riders to try to create more demand, or we can send bonuses to drivers to try to create more supply. And that money that is spent that comes out of the budget. And there's this planning process of deciding how much to spend on either side of the market.
So to create a plan like that, you need to have an idea of what you expect to happen if you don't spend the money, which is a forecast. So we're going to say, we're gonna make a forecast of how many drivers will show up next week, how many riders will show up next week, and then kind of forecast how much imbalance you're going to have and then imply some course of action, we should spend this much money on this particular thing.
So I worked on for a couple of years building a forecasting and planning system to solve that problem, which meant that we had to kind of integrate with all these systems to do forecasts of all the different riders and driver behaviors. And to forecast what would happen under different scenarios. If we spent more or less money on those different kinds of supply and demand levers, and then also to predict what would happen under different circumstances for different amounts of riders and drivers on the road, based on the historical data. So all those problems individually are super challenging. And then once you can solve them all, you can actually start to do something like a budget optimization. So start to think about, what's the optimal amount of these decision variables that we're going to spend?
This was a really fun and ambitious project and really core and fundamental to the business that Lyft operates in. And it really did rely on a lot of different skill sets, including, being able to forecast these things that are pretty challenging to forecast, they have a lot of seasonal patterns and in growth and different kinds of events that drive them. But also there's a causal inference aspect to that, which is another thing I really enjoyed working on, which is just, if we spend more money, what will happen?
I'm sure a big part of your job is just getting buy-in from your team to invest in such a complex model?
It's not even just buy-in as a one time thing, you have to kind of continually maintain trust, understanding and getting buy-in from all your stakeholders on projects. I talked about working really openly 20 minutes ago, that's exactly what you need to do on a project that is keep communicating the progress you're making, setbacks you're facing, and then continually educating people about what the benefits of what you're doing, will be when when it finally works. All those are kind of very soft skills, but they're really important to the success of a data science project.
So actually, this brings up a really good point, which is that what you described, definitely sounds like a highly data intensive project. And also sounds like you were in an environment where people understood that data is key to informing a lot of decisions in terms of how the product was taking shape. And of course, the data scientist has to do a good job in terms of earning trust of their team. But, I'm sure that the environments can vary from company to company, and not all teams or companies might think about data similarly. So, if you're a candidate, and you're interested in data science, is there a way for you to figure out how data science is viewed as a function internally?
That's a great question. It's pretty challenging to do that as an external candidate. And I've seen it on both sides, when I was thinking about joining the team that I ended up joining, it took me a really long time to build the confidence that the system I just described to you, it sounds complicated, and that people at the company would not expect it to work. And I was very worried about that, you want me to come and build this thing and work on this complicated thing - Do you really believe that you have the social capital internally, and the political capital internally to make something like that actually drive decision making?
And that was a big thing I had to resolve in my head before joining and then kind of on the other side of the marketplace, where I'm trying to hire people often, they're asking very good questions about what's the role going to entail that sometimes I can't answer - are we going to be able to do these things when I join your team, and sometimes you can't make promises about that. Every company is very different. And it's so hard to know in advance what you're getting into. Lyft is a place where there is a very science heavy input into the way the company runs, and if you see the power structure, when someone in science had good ideas they tended to get implemented or worked on. I thought Facebook was like that too. But I've heard about other companies where, there's an engineering or product first culture and the science can come second or third to those things. And, that's really common in practice.
So, I don't know how to know in advance, I've only worked at two places. So I can't claim to be able to read the tea leaves about what kind of places are going to be the best to work. But talking to the other data scientists at the company, and seeing the kinds of places that are able to effectively grow are probably good proxies, if they're hiring lots of smart people that you respect, then they're probably doing something right. And certainly, your individual hiring manager, the person who will manage you is probably the best single thing to know about in advance before you end up at a company because they're the person that's going to be able to get buy in for what you'll be working on. Or they've already gotten buy in do it. But those are good questions to ask is, to what extent do you believe that the company and the leadership are bought into what you want to do, or what makes you convinced that the thing we're going to work on is going to get the buy in order to continue to get resources to make it into products, and in hearing them explain to you how they ended up kind of coming to that conclusion probably would be a useful thing to ask as well.
Would you say that there are certain kinds of products that lend themselves more to being highly data driven?
That's a really excellent point. And I wish I had thought of it, which is that certain businesses are just more amenable to data driven and automated decisions. So, one framework I have for that is that if you think about data, the claim earlier that data science is helpful for improving decision making, you'd want to have a lot of decisions. The more decisions they have the better for data scientists who want to help them improve more decisions. So a stream of any decisions is important. So if you can think about the volume of decision making and then the individual importance of those decisions is another piece to it. And then how much incremental value does data add to the making of that decision. And then if you multiply all three of those things together, you might get something approximating. How much value could be created through applying data? But if you work in like fashion, you're the head of J Crew or something you're not going to use data to really tell you what clothes to make. But it might be useful in optimizing your supply chain and making sure that you have enough pairs of pants in the store. So I think that's one thing that's been a little interesting to me as I talk to people in other fields is that they're often in parts of businesses that are really data focused you wouldn't think of at first blush. And there's lots of companies that don't look like tech companies, but then have sort of nascent data and engineering focuses in areas that you might not think about. So that's kind of worth being open minded about what kinds of problems could exist.
Do you recommend any other questions that candidates should be asking during the interview process?
I wish I had more intelligent things to say here, because I honestly haven't been on that side of the job market enough times to know to have a good sample. Data science is a role where the most common frustration that I see is people are not given enough scope. So they work on things that feel too small for them or aren't going to be impactful enough to help them grow their careers, or aren’t intellectually stimulating enough. So things that aren't challenging are helping them continue to learn on the job, which is a really important part of being a data scientist, because the field is not static, you have to continually invest in learning. And if you're not given the opportunity to do that, I think people feel their careers are kind of stagnant. So both of those things are hard to know in advance, but they're ultimately going to be determined by the manager you will have - does the manager have a meaty enough role for you? Is there enough scope for what you're going to be doing to grow over time? And that would be something I would ask about is what's this role going to look like in a year or two? And what's the total value I can add to the business that I'm joining? And then the learning opportunities thing is kind of, are there challenges? Are there unsolved problems, you're gonna need somebody to work on? Or is there scope for me to learn on the job and to try new things that I'm not currently an expert in? Are you going to create the opportunities and time for me to be able to do that? Or am I going to be just totally overwhelmed by what we live with and call keeping the lights on? And I think those two things are probably the biggest things that I would ask about, because I think those are the reasons people end up leaving jobs is companies - they either don't have a good manager, or they don't have enough scope where they are. They're not learning what they want.
Would you recommend any resources for candidates who are interested either in learning more about data science as a field, or for interview prep?
I'm really reluctant to recommend any particular one thing, if you feel in line with the way I've answered a lot of questions so far is that there's a learning process to both get a job and to be good at a job. They're somewhat correlated, but not the same thing. The advice I always give people is to get interested in what they need to be good at, because if you can't be interested in it, you're never going to be motivated to do it. So if I said, go read this book, and you weren't actually intrinsically motivated to read it, it wouldn’t probably work in the first place. So the primary factor is whether you think it's interesting. And you have to be really interested in data science, which means finding stuff that's interesting to you.
And what I would recommend is there's a variety of ways to do that. There's lots of great YouTube content, you can be watching videos, or you can be reading books, or you can be on Twitter and chatting with people or you can be working on projects that are kind of personal to you and things that you find interesting. But if you're not passionate about learning and finding yourself doing those things on your own, without someone telling you to do it, just with the without needing to get a job, I think it's going to be challenging to keep growing your career. Because it's all about energy and managing how much energy you're able to put into that stuff. And the energy really comes from having like a deep curiosity for the field that you want to get into.
Note from LED: Here is a highly rated data science course on Coursera that's offered by IBM, that might be worth checking out! We had another guest on LED, Yash Kotresh, who was a Data Scientist at WalmartLabs at the time, and recommended these two courses on Coursera:
Course for beginners on Machine Learning, offered by Andrew Ng
Thanks, Sean! What's the best way to reach out to you if someone wants to?
DMS on Twitter or on LinkedIn are totally fine :)
If you have any feedback to share, or if you have any questions for Sean, drop us an email at firstname.lastname@example.org or tweet at us @LED_Curator