Uber Data Scientist Interview Guide

The Uber Data Scientist process usually opens with a recruiter conversation and moves into technical screens that mix SQL (joins, window functions, lag/lead, dense rank, CTEs) with basic Python, such as sorting a list or writing a simple function. Experimentation runs through almost every stage. Candidates were asked how long to run a test, how to think about sample size, what to look for in results, what to do when an experiment fails, and how to verify findings. Case studies lean on Uber's own marketplace. Examples include Uber Eats order or driver assignment framed as an optimization problem with objectives and constraints, the effect of batching food orders, and whether showing driver ETA one minute less changes behavior. Another was estimating the price elasticity of delivery fees from observational data, where the interviewer pushed hard on endogeneity. Causal inference also comes up directly, with questions on when to use synthetic control, A/B testing, or difference-in-differences.

Longer loops add a hiring-manager interview, a bar-raiser behavioral round, and sometimes a dedicated experimentation or econometrics case. One candidate was told the bar raiser checks alignment with Uber's stated values across every story. Frame conflict and stakeholder examples around business and customer impact. Some loops also include LeetCode-like coding or probability puzzles, such as a sequence of dice expected-value questions in an intern case round. Keep core coding and probability sharp even though the role is analytics-oriented. Outcomes in these reports skew toward rejections, including one that arrived two days after a SQL screen and an A/B testing case round with little actionable feedback. Case topics differ by team, so treat any single prompt as illustrative of Uber's style rather than a fixed syllabus.