Data Scientist interview
at Meta.

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 used data to influence a product decision that a partner team or PM initially disagreed with.
2. 02 Describe a project where you had to move fast and ship an analysis with incomplete or messy data, and how you handled the trade-off.
3. 03 Tell me about a time your analysis surfaced an unexpected insight that changed the direction of a feature.
4. 04 Describe a situation where you had to push back on a stakeholder who wanted a metric or experiment interpreted in a way the data didn't support.

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

1. 01 What product signals would you use to determine who a Facebook user's 'best friend' is in order to prioritize their content in Feed?
2. 02 How would you design an A/B test to measure whether a new Reels autoplay behavior improves engagement without cannibalizing other surfaces?
3. 03 Write a SQL query using window functions to compute 7-day retention for users who first posted a Story in a given week.
4. 04 If daily active users on Facebook Marketplace dropped 5% week-over-week, how would you investigate root cause and decide whether it's real?

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

1. 01 How would you define and instrument the success metrics for Facebook Groups, including a guardrail metric for community health?
2. 02 Design the metrics and experimentation framework you'd use to evaluate a change to the News Feed ranking algorithm.
3. 03 How would you measure the overall health and success of Facebook Marketplace as a two-sided marketplace?
4. 04 Design the metric framework to detect and explain a sudden 15% drop in Instagram Stories engagement.