High-Level Design rounds at Paytm lean toward real fintech-scale problems — risk/fraud scoring, booking-style concurrency, and database schema trade-offs — rather than generic textbook prompts. Interviewers reward a structured approach: requirements first, then schema and data model, then how the design holds up under load.
Problem 1Design a Risk Management Service
Design a service that scores and manages fraud/risk for transactions flowing through a payments platform, at a scale where the database schema and query patterns become the main bottleneck.
Requirements to clarify first
- Is the score needed synchronously before the payment is authorised (a hard latency budget of, say, under 100 ms) or can some checks run after the fact and trigger a reversal?
- What signals feed a decision — per-user velocity (transactions in the last 10 minutes / 24 hours), device and location history, merchant risk, block-lists — and which of them must be exact versus approximate?
- Who consumes the output: only the payment flow (approve / decline / step-up OTP), or also an analyst review queue and a rules team that changes thresholds without a deploy?
More system design prompts Paytm has asked recently
The full architecture, schema design and scaling discussion for these are part of the Placement-Ready PYQ Kit.
1 more system design prompt with a full architecture writeup in the full Placement-Ready PYQ Kit