Data Scientist interview
at Uber.

Tell-me-about-a-time stories on ownership, conflict, and impact — scored against this company's real values.

1. 01 Tell me about a time you acted like an owner and drove an analysis end-to-end across product, engineering, and operations partners without being asked.
2. 02 Describe a situation where you had to influence a product manager or city operations team to change a decision using data when they initially disagreed with you.
3. 03 Give an example of a time you persevered through messy, incomplete marketplace data to still deliver a trustworthy recommendation under a tight deadline.
4. 04 Tell me about a time your experiment or model result was counterintuitive and you had to convince stakeholders to make a bold bet on it.

The hard screen for the craft itself — talked through out loud, not whiteboarded in your head.

1. 01 You roll out a new feature and trip completion rate drops in three cities but rises in others; how do you design an A/B test and decide whether to ship it?
2. 02 Walk me through how you would detect and correct for network interference and spillover effects when running a marketplace experiment that affects both riders and drivers.
3. 03 Given a trips table with users, cities, and timestamps, write SQL using window functions to compute the average delay between a user's signup and their second completed ride.
4. 04 How would you measure whether a change to surge pricing improved marketplace health rather than just short-term revenue?

Open-ended design of the systems this company actually runs, with the interviewer probing your tradeoffs.

1. 01 Design an experimentation platform for Uber that can run thousands of concurrent rider- and driver-side A/B tests while controlling for interference between treatment and control.
2. 02 Design the metrics and monitoring pipeline that would tell you in near-real-time whether trip completion is degrading in a specific city.
3. 03 Design a system to estimate the causal impact of surge pricing on driver supply and rider conversion in a given market.
4. 04 Design a forecasting system that predicts city-level rider demand by hour so operations can position driver incentives ahead of time.