How Lyft Data Scientists Have Worked Remotely During the Pandemic
COVID-19 has brought many changes to daily life, and the routines of our Lyft Data Scientists have been no exception. We sat down with Andy, Faten, Garrett, and Yibei for their takes on working at Lyft during the pandemic.
Tell us a little bit about what you do at Lyft, and what a typical day looks like?
I’ve been a data scientist at Lyft for around one year, around the same time my son was born! I work on the request flow part of the rider app, basically which ride-types we recommend to the rider in certain situations.
In caring for my child, sleep is sometimes hard to come by, but both Lyft and my manager handle this very gracefully and are understanding of my situation.
I usually get up early to put my son into daycare. I highly value these conscious moments before dropping him off — when he is still very alert and smiley. After coming back home, there is still half an hour for some exercise that I selfishly claim to combat screen fatigue. I usually sign on around 9AM with no meetings for 1–2 hours. I check emails, Slack, prepare for presentations, evaluate overnight models, or focus on some more fun projects during this time.
In the early afternoon I have a couple of 1–1s and team meetings which require preparing and documentation. Ad-hoc conversations with stakeholders around data or business issues also tend to happen during this time. The late afternoon is when I work on long-term projects. For me it’s good focus time. I pick up my kid around 4:45–5PM, go for a walk with him, and tuck him into bed.
I generally wrap up with a couple of hours of coding during the evening — there are usually no distractions then so it’s a great time for heads down work. That generally caps off my day at Lyft, and I follow this schedule during the week; weekends are work-free time and that is certainly something I like about Lyft’s culture.
I work on the Revenue Operations Science team, where we create tools and analyses to help stakeholders make operational decisions rather than work on a consumer-facing product.
A typical day of mine starts with a lot of analysis, usually for tool building, debugging, or upcoming operational decisions. I also often put together slide decks for stakeholders and cross-functional teams, meet with engineers and other collaborators to share project progress, and provide guidance on the tools I’ve developed.
This is my first job in industry. Before Lyft, I wrapped up my Ph.D. in Neuroscience and pivoted into data science last summer. There is a high meeting load, but I can see their value. This definitely requires good time management skills and active planning of heads-down time to ensure that I’m pushing projects forward.
I am a scientist with the Lyft Transit Bike and Scooters line of business, where I support Lyft’s hardware team.
What I do on a daily basis varies depending on where our team is in the hardware product development cycle. Are we launching a new product? Are we planning the next-generation hardware? Analysis, modeling, presenting, meeting with stakeholders: it just depends on the day and what stage we are at.
In addition, our team is unique in that a good amount of our time can be spent doing hardware engineering. Most of our team has some background in hardware, and I did a Ph.D. in experimental physics before becoming a data scientist. If we have an issue that comes up in testing or in the field, we’ll work to debug it and help change the design. As such, when we’re first launching a product or discovering an issue, data scientists are very valuable to the engineers because we can give them data about what’s actually happening with hardware in the field. We will also help in the initial hardware design process by applying driven frameworks to a hardware product’s business case, product requirements, and engineering architecture.
The hardware team consists of 50+ engineers in specialty disciplines that all work together, and we try to bring discussions to resolution with data, whether on the product side or the engineering side. There’s a lot of talking with cross-functional partners in addition to doing analyses, building dashboards, and building models.
I am a part of the Trip Experience group. From request to drop-off, we look at cancellation problems across the whole rider journey.
Last winter, I made a decision to work remotely from Hawaii, to get a respite from the cold of New York. The COVID situation at the time was better managed, and there were a lot more things to explore. I’ll speak to my typical day in the context of my time there:
There was a 5-hour time difference between Hawaii and NYC, meaning quite early mornings. I learned to adjust my time to load a lot of meetings towards the afternoon in Eastern time, and colleagues were adaptable to these changes. I also did my ad-hoc analyses and responded to slack channels in the mornings, and used the afternoons as the focus time for activities such as deep dives, experimentation analysis, and strategic thinking.
After work, I was able to have dinner with friends outside, and there was always a lot to do during the weekends. I came back to NYC in early 2021, but I loved my time in Hawaii.
How do you feel about working remotely during COVID? How does COVID change your daily work?
Working from home this past year has added a significant amount of flexibility, which has been extremely helpful as a new parent. I also believe data science is not necessarily a 9-to-5 job — sometimes you work on a project, do something else, take a break then come back and work some more.
I’ve been very happy with my work situation so far — I feel that a lot more gets done in my current role than my previous positions, and every interaction I’ve had with colleagues has been very structured and well organized.
To share a fun story, due to COVID, the onsite interview was done remotely, and between interviews I was very nervous so I did some pacing through my garden to destress. I stepped on a wasp — barefoot — which was a bit unfortunate. Thankfully nobody noticed and I still got the offer; a small price to pay.
I started this job remotely, so I don’t have a sense of what it was like before COVID. I feel that I’ve been able to be efficient while working alone, as I can focus on projects for hours at a time, and less time is lost context switching, but it can take longer to get questions answered and it is harder to onboard remotely. The biggest pain point has been feeling disconnected from coworkers, as we are not all in the same physical space with each other. Hanging out with team members virtually does help with this!
I have two cats, and working from home with them is pretty fun. Sometimes my cats jump onto my desk and walk over my keyboard, sending an unintelligent message over Slack, or editing my SQL query. They are a great source of stress relief.
Our team has managed well remotely, but there are certain tasks that are best served in-office, so we go in from time to time. Hands on time with hardware is really important.
As I used to commute with Caltrain, the first thing I had to do was convince my wife that I’d be safe to commute in. Another difficulty in COVID times is that people tend to schedule a lot more meetings virtually, so I have to figure out how I’m going to get to the lab, which meetings I can postpone, and which ones I’ll take from the lab (which can get loud with machinery). Lyft has strong safety protocols once in the office, but getting there safely has been the bigger challenge.
As a COVID-times anecdote, one day I went into the office to help build bikes for an internal beta. We set up a 20-step assembly line, and my goal for the day was to go from steps 3–9. I did less than half of that because every time somebody came in that I hadn’t seen in person in about a year, we’d catch up a little bit.
I am still relatively new to the company, as I joined five months before COVID happened. I switched teams during that five months, so I was still in an onboarding period when COVID started. Between COVID, joining a new team, and covering two pods within Trip XP, work was challenging in the beginning. When COVID hit, we had the incremental difficult task of positioning the business for a successful rebound. However, I felt extremely supported by my manager and the other members of my team.
In general, I definitely miss having face time, as when people meet remotely, people tend to be very busy, jumping in and out to their next meetings and adhering to the agenda.
In Hawaii, there were many things to explore over the weekend, and less COVID, so it felt safer to go outside. I found that overall, moving to Hawaii made me work more efficiently during the daytime, because my stress level decreased.
Q) What tools/skills or past experiences help you succeed as a data scientist at Lyft?
One thing that I think really helped me is being curious and excited about what I am working on. This helps me stay focused, and drives me to ask good questions that further my understanding. Another thing I try to do is not jump to conclusions; as a data scientist it’s important to always be able to back up your arguments with data and that’s something I try to evangelize.
In my past experience as a physicist I worked in small teams in various labs. This taught me that it’s super important to proactively help as much as you can, because it creates a kind environment that everyone wants to continue working in — especially if it’s hard sometimes. It seems that at Lyft I have found the same helpfulness being propagated, which creates an extremely welcoming environment.
The biggest skills that carried over from my previous experience would be forming data-driven hypotheses and turning an open-ended problem into an actionable project with explicit success criteria.
Another skill that has been helpful is the ability to communicate science to lay audiences; my volunteer experience giving scientific talks to a general audience in graduate school helped me a lot.
I think people in data science tend to be very focused on using methods to answer a problem. However, my impression about what’s most valuable for data scientists is to focus on the problem and to provide a solid answer, no matter what the data looks like or the tools at your disposal. This includes even documenting our opinion when might not have sufficient data, from the viewpoint of how we approach it as a data scientist. A Data Scientist with partial data or a framework alone can bring a new perspective to many conversations and decisions.
First, leave time to breathe, and get to know your coworkers on a more personal level. Chat about non-work-related stuff! We tend to be very efficient and technical in terms of solving the meeting agenda items, but sometimes we might overlook the importance of relationship and trust-building. And that also means that when we have different opinions on the project, we should feel free to disagree and commit.
Second, make sure you’re keeping a good balance between work and life. As we’re all working from home, there is a tendency to fill commute and entertainment hours with more work hours. If you don’t remind yourself to keep these hours to yourself, you can evangelize unhealthy behaviors. The day-to-day can become repetitive, so find something that breaks you out of that cycle, gives you something to look forward to, and in turn makes you more efficient.
Q4) Any parting words or advice that you can give to an aspiring Lyft data scientist?
For the interview, I think it’s important to really nail what are some of the key metrics, specifically for the transportation sector. It’s also a good idea to brush up on statistics before the interview!
Based on what I have observed so far, know your SQL and python, know your stats, and practice presenting clearly and succinctly. It will also help if you are able to adapt to the fast pace and constant changes at Lyft by turning projects around quickly while maintaining high quality work, and then iterating on them to get to a more final, thorough product.
My advice for interview preparation is presenting everything in the context of business decisions, keeping in mind questions like ‘why does this matter for the company and customers’, and ‘what is the incremental value of doing this’.
I would encourage people to think about the problems and the framework, and use your business and engineering judgement. Think through how you can contribute even when you don’t have great data, thinking less about the tools and more about the problems that need to be solved.
The Science org at Lyft has been spending a lot of time building a strong community among scientists. Efforts originating from the Diversity and Inclusion Group and Science Learning and Development Councils have been very helpful in fostering a sense of belonging and inclusion.
If you are interested in joining our incredible team of Data Scientists, check out our Careers page!
A huge thanks to the following for their great contributions to this post (in alphabetical order): Ava Li, Carol Zong, Farshad Majzoubi, Michael Zhou, Simran Mirchandani, Tim Xu, Xin Wang