Lyft Data Scientist interview questions

It helps to know someone in the company to fast track you to a recruiter and interview. After the phone screen about the role/potential fit, they send you a dataset to analyze in a 24 hour window. If they like your submission, they ask you to come in for a presentation of your work, in which they grill you on the assumptions you made in your analysis.

Overall straightforward interview -- but, make sure to know your ride-sharing metrics and how experiments look like at a ride-sharing company or else you are screwed. Interview process took 1 month. FYI, in the Lyft context "Data scientist" = product analyst, for the most part Process: (1) Got referral to start process (2) Phone screen w/ current data scientist. Go in depth with a past project, talk through the math of hypothesis testing ("why use a t-distribution in this scenario?"), talk through success metrics for a ride-sharing business case (3) Take home: there are a set of questions you have to answer with a simple ride-sharing dataset (how to measure churn, set up experiment for recommendation to reduce churn) create presentation for on-site. (4) On-site: 5 interviews. (1) Presentation of take home. My advice is to evaluate your definition of churn with a false positive rate and know about clever experiment set-up (i.e. splitting your experiment across multiple cities, etc.) (2) SQL test (3) business case -- be sure to know the unit economics metrics related to ride-sharing. Be able to answer questions like "what data told us to create Lyft shared rides?" (answer: overlapping routes in map data) (4) stats & probability: had to answer the hypothesis testing classic "Coin got x heads during y flips. How can we test if this is a fair coin" and then pivoted to an ride-sharing experimentation question (5) Core values interview: really cool to hear about how Lyft thinks about success, culture, and evaluating the performance of data science ICs I didn't get an offer. Basically, if I were do this again I would do the metric prep that I did and then ALSO grab coffee with a friend working in ride-sharing data science to make sure I understand the metrics that matter.