Diving into Data Science

Meet Alya, Mike, and Troy! They are some of our talented Data Scientists who joined the DS space through less traditional paths. In this Q&A, they open up about how they got to this point in their careers, and share some advice to others looking to make the switch.

Alya Abbott

Alya Abbott

Q) How did you get here? How do you feel this path equipped you with the skills necessary to succeed in Data Science?

Until I moved to California from Boston in 2014, I was unfamiliar with the term “Data Science.” I picked up the skills of a data scientist by following my interests and learning as I went along, and was lucky to find that those skills were in high demand.

I did my PhD in theoretical linguistics, intending to go into academia. My thesis was on syntactic phenomena in Uyghur, Russian and Faroese, and had nothing to do with Natural Language Processing (NLP). However, linguistics is a tiny field with a very tough job market. I ended up working at MIT Lincoln Lab after graduation, doing applied research in crowdsourcing, NLP and machine learning. Lincoln Lab hires based on the ability and willingness candidates to figure things out, which a PhD is great preparation for. As I didn’t have a lot of relevant experience to draw on while at Lincoln Lab, I was constantly learning and developing. While relevant prior experience in my work was minimal, it was an amazing place to learn.

After a couple of years at MIT Lincoln Lab, I moved from Boston to California and started looking around for what to do next. This is when I realized that in the Bay Area, what I had been doing at Lincoln Lab was called data science (in 2014, the term hadn’t made it to Boston yet!). I took a job as a data scientist, gradually transitioning from an individual contributor (IC) to management. I feel incredibly lucky to have found an area with so many fascinating problems and opportunities.

Q) Are there things that surprised you or really challenged you once you made the transition?

What surprised me most when I moved into industry was just how much collaborative effort is required in order to ship a product. Even something that seems like a “small” feature with a data science model at its core can require contributions from Product, Engineering and Design in addition to Data Science, as well as collaboration with other teams. This is very different from the research work I saw in my PhD and at Lincoln Lab, where one person — or a small handful of technical experts — would complete a project end-to-end. It gave me a sense for why companies emphasize communication skills so much for highly technical roles.

Q) What about our Data Science organization drew you to Lyft?

Even before I started interviewing at Lyft, I had heard about the collaborative, mission-driven culture at the company, and the interview process was a great way to learn more. The focus on making a positive impact attracts people who are passionate about what they do; it’s fun being at a place where people work hard and help each other make things happen. From the interviews, I learned about the huge variety of data science problems that Lyft tackles. It’s been a great place to work, with so many smart people from a variety of backgrounds to learn from.

Q) Lastly, what is the best advice you could give to someone hoping to pivot into Data Science?

Find ways to get both theoretical and practical experience. Figure out a problem that excites you that could be solved with data science tools and try to solve it. Actually make it work, but also read enough theory to understand why it’s working — what’s happening under the hood? Why did some approaches you tried fail?

Throughout my career, the main resource I’ve turned to is the people around me. Whenever you’re trying to do something for the first time and are feeling unsure, talk to people who’ve done it before! This applies to technical work, but also to non-technical skills like project/time management or navigating relationships to get something done at the company you’re at. People are generally very generous with sharing their experience; even if you don’t know them well, just ask!

Mike Frumin

Mike Frumin

Q) How did you get here? How do you feel this path equipped you with the skills necessary to succeed in Data Science?

My path to working as a data scientist at Lyft started with my fascination with using and improving public transit. The data science aspect came later. I was born, raised, and currently live in, Brooklyn. I started commuting by subway in high school, and have never owned a car (nor do I plan to, despite having two small children). After early career experiments around different applications of software engineering, I sought a way to marry my technical skills with personal priorities around dense cities and the many economic, social, and cultural benefits they bring.

I honed in on Public Transportation (known as Transit at Lyft) as an industry to focus on, found the Transportation program at MIT, and was able to dust off the mathy side of my brain while doing data-driven research with large transit agencies in New York and London. I ended up working in Product and Engineering for NYC’s real-time bus tracking system and then on the largest bikeshare systems in America. That is, not much data science per se.

When Lyft was opening its office in NYC, I was invited to interview to build and lead the Data Science team there as a function of (I believe) my academic training, deep transportation industry background, and cross-functional leadership experience. The work at Lyft – especially as Lyft has gone all-in on transit, bikes, and scooters – has proven to be a perfect way to leverage those skills. What I learned in my study of the rich literature around ‘traditional’ transportation systems modeling turns out to apply directly to rideshare and micromobility. My experience in previous roles helps me collaborate effectively with our cross-functional partners, which is a key requirement for any Data Scientist. Even my time at a century-old public sector infrastructure and operations bureaucracy helped me learn how to navigate large organizations, to be deeply empathetic to people with very different responsibilities, and to communicate my technical work in terms others can easily understand.

Q) Are there things that surprised you or really challenged you once you made the transition?

While it seems obvious in retrospect, I was surprised how important the analytical work to drive human decision-making at Lyft was. I had imagined that the lion’s share of Data Science work was to build the automated models in the product, but I have come to really appreciate how critical Data Science can be for driving strategy, product roadmap, and tactical business decision-making.

Q) What about our Data Science organization drew you to Lyft?

Before the Data Science organization, I was first attracted to Lyft because it shares my personal mission around transportation for livable cities. Then, the fact that Lyft is pursuing these goals through the use of technology and data played to my technical strengths. I felt that I had a tremendous amount to learn from the company, while being able to provide a unique perspective given my background in other transportation contexts.

Coming to Lyft from outside the broader Tech industry and as a not-so-young person (with kids), I was a little afraid that I might be an outlier on the team. However, the very first piece of content in day 1 of onboarding highlighted how many of the 90 people in the onboarding class talked about their children. It immediately made me feel right at home.

Q) Lastly, what is the best advice you could give to someone hoping to pivot into Data Science?

At Lyft we expect Data Scientists to “continuously earn a seat at the table through thought leadership and deep collaboration.” We are knowledge-workers in the purest form (i.e. creating knowledge), so none of the work matters if our colleagues don’t consume, understand, and believe it. This is about both which work you do — which problems are you going after to drive the most benefit — and also about how you communicate the results. Double down on thinking about your audience, understanding their needs, and delivering your work and insights in a way that speaks to them.

Troy Shu

Troy Shu

Q) How did you get here? How do you feel this path equipped you with the skills necessary to succeed in Data Science?

My path to Data Science was a little circuitous but I think it prepared me well. After studying Computer Science and Economics in undergrad, I worked at a quant hedge fund where we used Python pandas every day. I didn’t know it at the time but a whole new field called Data Science would soon be created around such tools.

After a year, I left finance to work as a backend software engineer at a startup. I learned a lot about software development processes there. Throughout this time I built (and still build) software tools, which is where I learned about the importance of analytics in building software that users love.

While I enjoyed building software, I also liked finding insights in data. I found I wanted to influence — and even make — higher level product or strategic decisions, so I left my software engineering job and spent a year “exploring.” During that year, I freelanced as a software engineer with data chops, tried to start a company with friends (twice, neither worked out), and explored both Product Management and Data Science job openings. By now Data Science had exploded in popularity. I decided that Data Science was a good fit for me, and here I am!

Q) Are there things that surprised you or really challenged you once you made the transition?

When I first transitioned into Data Science, I was most surprised by how much Data Scientists (at least product-oriented ones) rely on SQL and A/B testing on the job. For example, while I was familiar with programming languages like Python and had used pandas a lot, I never had to learn SQL before becoming a Data Scientist. SQL wasn’t taught in my Computer Science classes in college. Looking back, all the SQL practice I get every day has really shown me how being fluent in SQL helps you become much better at both thinking about and wrangling data.

Even though I minored in statistics and understood the theory behind A/B testing, it was nothing like on-the-job experience and learning from my coworkers about the important details of actually applying A/B tests to a product. Examples include the need to define both primary and secondary or “guardrail” metrics, doing power analyses where sampling occurs over time to estimate A/B test length, and why changing variant allocations or shocks like marketing campaigns in the middle of an A/B test can bias results.

Q) What about our Data Science organization drew you to Lyft?

I really liked how data-driven Lyft seemed to be, given the Data Science blog posts I read from Lyft and also what I learned from my Lyft interviews and interviewers. Data Science is treated as a critical function here. We have internal tools and even teams dedicated to democratizing data and data-driven decisions, and a culture that not only sees data as something to report on but also as a competitive advantage, core to both the product and to our strategic decisions.

What put Lyft above other companies in my mind was the inclusive and diverse environment that the company cultivates through action. For example, this blog, and this specific post, is a manifestation of just that because it showcases people from diverse backgrounds — in not only gender, ethnicity, age, etc. but also life paths — and their unique insights.

Q) Lastly, what is the best advice you could give to someone hoping to pivot into Data Science?

My advice for someone hoping to pivot into Data Science is to prepare early and well. Being a Data Scientist requires a lot of different kinds of skills, both technical and non-technical. Understand what Data Scientists do and develop and practice those skills. Demonstrate those skills through your current day job. And if that’s not possible, create separate opportunities for yourself to showcase those skills — like through doing side projects — and put them on your resume. Make it easy for your future employer to have confidence that you can do the job well.

Of course, interview performance is critical too, so prepare well for that. Reach out to data scientists and talk to them about their experiences and the broader interview structure. Brush up on relevant topics and find some practice problems online. You can even record yourself explaining your answers and giving responses, then make adjustments to how you want to present your problem solving process, your background, etc.

We hope these team members’ stories inspire you to pursue Data Science if that is where your passion lies, irrespective of your background!

If you are interested in joining our incredible team of Data Scientists, check out our careers page!

A huge thanks to the following people for their great contributions to this post (in alphabetical order): Eric Smith, Polly Peterson, Sarah Morse, Simi Mirchandani


Diving into Data Science was originally published in Lyft Engineering on Medium, where people are continuing the conversation by highlighting and responding to this story.

Source: Lyft

Leave a Reply

Your email address will not be published.


*