ML System Design Fundamentals — Student Lab

Case study: Fraud detection for card-not-present transactions.

This notebook is a structured design doc exercise. Fill in the templates with concise bullets.

Prompt (the interview question)

Design an ML system to detect fraudulent card-not-present transactions in real time.

You must address: requirements, data, modeling, serving, monitoring, and iteration.

0 — Problem framing

1 — Requirements (SLOs + constraints)

1 — Requirements (SLOs + constraints)

2 — Metrics (business + model + guardrails)

3 — Data + labeling plan

4 — Modeling plan (baselines → iterations)

5 — Serving architecture (online + batch)

6 — Monitoring + iteration loop

7 — Final: tradeoffs (2 required)