Went through the Amex DS loop earlier this year for a mid-level role on their risk modeling team in NYC. Sharing notes because I wish I'd had something like this beforehand.
The process had four rounds:
Phone screen (recruiter): 30 min, standard background stuff. They wanted to know if you'd worked with financial data before. I hadn't, and it didn't disqualify me, but they did ask follow-ups about how I handle regulated data environments.
Technical screen (30-45 min with a DS on the team): Two SQL questions, both multi-table with window functions. One was a rolling 30-day customer spend calculation. The second was about finding outlier merchants using a self-join pattern. Genuinely intermediate difficulty, not "list all customers" stuff. They also asked one stats question: explain A/B test power to a non-technical stakeholder. Clear framing, explain the tradeoffs.
Case round: This surprised me. It was a semi-structured case, not a consulting McKinsey case, but more like: here's a business problem (churn prediction for charge card holders), walk us through how you'd approach it. They cared more about problem decomposition and feature thinking than modeling specifics. I mentioned XGBoost and they didn't press on tuning details at all. They wanted to know how I'd validate and monitor the model post-deploy.
Final loop (3 rounds, same day): One more technical (Python, Pandas, no LC-style coding), one behavioral with a senior manager (heavily STAR-based, lots of "tell me about a time you disagreed with a stakeholder"), one values/leadership round with someone from their leadership pipeline.
Comp I was offered: around $160k base, $25k sign, 15% annual bonus target for mid-level DS in NYC 2026. Felt reasonable for fintech, not FAANG-competitive.
One thing nobody told me: Amex has strong opinions about "enterprise thinking." If you only have startup DS experience, be ready to talk about how you'd scale work across teams, not just ship a notebook.