API design as a product, webhook reliability, and the correctness bar that payments infrastructure demands.
Razorpay sells an API, which changes what the interview values. API design stops being a sub-question inside system design and becomes a round in its own right: resource modelling, idempotency semantics, versioning without breaking existing integrations, error contracts a developer can actually handle, and pagination that stays stable while data changes underneath.
Webhooks are the second distinctive area. Delivering an event to a merchant endpoint that may be down, slow, or returning a success while silently failing is a genuinely hard reliability problem. Retry with backoff, signature verification, ordering guarantees and an explicit replay mechanism are the things interviewers listen for.
45-60 min
Medium difficulty with emphasis on clean, correct implementation.
60-90 min
Design a developer-facing API and model it in code: resources, errors, versioning.
60 min
Payment flow, webhook delivery, reconciliation and failure handling.
45 min
Ownership, on-call, and decisions made under correctness constraints.
Because the API is the product. A poorly modelled resource or an unstable error contract is a customer-facing defect that cannot be quietly refactored away, so the interview tests whether you design for external consumers who will never read your code.
At-least-once delivery with exponential backoff, signature verification so merchants can trust the payload, what ordering you do and do not guarantee, dead-letter handling, and an explicit replay path for events a merchant missed while their endpoint was down.
Both demand payment correctness. Razorpay adds a developer-experience dimension - API ergonomics, versioning, webhook contracts - because its customers are engineers integrating against it rather than consumers using an app.
Practise the rounds Razorpay actually runs, and get scored on where you stand.