Data Scientist interview
at Spotify.

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 had to explain a complex statistical result to a non-technical product owner so that an intelligent person with no data science background could act on it.
2. 02 Describe a situation where you used your autonomy within a squad to pick a data approach the rest of the team initially disagreed with.
3. 03 Walk me through a time an experiment you ran produced a result that contradicted a strongly held assumption, and how you handled the pushback.
4. 04 Give an example of when you had to balance moving fast with a bias for action against the rigor of a fully powered experiment.

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

1. 01 How would you design and analyze an A/B test to evaluate whether a new Discover Weekly ranking model increases listening time, and which primary metric would you choose?
2. 02 Given user listening event logs, write a SQL query to compute weekly retention by signup cohort.
3. 03 When running an experiment on autoplay is infeasible, how would you use causal inference methods like difference-in-differences or instrumental variables to estimate its effect on engagement?
4. 04 How would you detect and correct for novelty effects and network interference when testing a new social sharing feature across the user base?

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

1. 01 Design the experimentation platform Spotify would use to run thousands of concurrent A/B tests across squads without metric collisions.
2. 02 Design a metrics pipeline that computes daily active users, stream counts, and skip rate for the company's product analytics dashboards.
3. 03 Design an offline evaluation framework to compare candidate Discover Weekly recommendation models before they reach a live A/B test.
4. 04 Design a churn-prediction system that flags Premium subscribers at risk of cancelling and feeds retention campaigns.