Interviewed at Chime for a data scientist role on their risk/fraud team in Q1 2026. DS interviews vary a lot by company so here's the actual breakdown.
The OA: SQL-heavy, two problems. Window functions, a multi-table join with an aggregation condition. Medium difficulty. Nothing I hadn't seen on LeetCode or StrataScratch. Time limit was 75 minutes and I had time to check my work.
Technical phone screen: One hour with a data scientist from the team, not a recruiter. Split into two halves. First half was SQL live coding: they shared a schema on screen and asked me to query it. Schema was clearly inspired by their actual data model (users, transactions, flags). I had to write a query to find users who had three or more failed transactions in a 30-day window. Sliding window problem.
Second half was stats/probability: A/B test interpretation, and then a question about class imbalance in fraud detection models. Both are core DS interview categories and I'd been expecting them. If you're interviewing for risk or fraud at any fintech, make sure you can explain precision/recall tradeoffs without referencing Wikipedia.
The case: Came in a separate round. Business scenario: how would you set up an experiment to test whether a new SpotMe limit increase affects user behavior. They wanted: hypothesis, metric choice, power analysis considerations, potential confounders. No right answer but they push on your thinking.
Behavioral round: Standard, mission-oriented like the rest of the loop. Why Chime, talk about a model you shipped, what happened when a stakeholder disagreed with your recommendation.
No ML system design in my loop but I've heard it comes up for ML-DS hybrid roles. For pure analytics DS it was SQL/stats/case heavy throughout.
Offer came back with a 160k base for the SF role plus RSUs. Felt reasonable for the level.