Razorpay Interview Questions 2026

Tip: Razorpay builds payment gateway, banking stack (RazorpayX), and lending products. Engineering domains: distributed systems for payment processing, financial data modelling, high-reliability backend (99.99% uptime SLAs), compliance and security (PCI DSS, RBI norms). Show familiarity with these constraints.

Tip: Pipeline: API gateway → idempotency layer (Redis dedup by payment_id) → payment service → bank gateway adapter (async fan-out). State machine: CREATED → PROCESSING → AUTHORIZED → CAPTURED/FAILED. Async callbacks from banks via webhooks, retry with exponential backoff. Message queue (Kafka) for event sourcing the state transitions.

Tip: Idempotency: calling the same operation multiple times has the same effect as calling it once. In payments: a network retry must not double-charge. Implementation: client sends a unique idempotency key; server stores the response against the key for 24h. Second request with same key returns cached response without re-processing.

Tip: Luhn: from right, double every second digit; if doubled value > 9, subtract 9. Sum all digits. Valid if total % 10 == 0. This is a real input sanitisation step in payment SDKs. Razorpay's interview tests both the algorithm and your code's readability.

Tip: Two-phase commit (2PC): Phase 1 (Prepare) — coordinator asks all participants to vote yes/no. Phase 2 (Commit) — if all voted yes, commit; else abort. Used in Razorpay when both the merchant account credit and the customer account debit must succeed atomically. Risk: coordinator failure between phases creates blocking.

Tip: At-least-once: persist event to DB first, then dispatch. On success, mark delivered. On failure, retry with exponential backoff (1s, 2s, 4s… up to 24h). Dead letter queue after N retries. Consumer must be idempotent (Razorpay includes event_id in payloads). Dashboard for manual retry.

Tip: Razorpay's SLA is 99.99% uptime. Show: you designed for failure (timeouts, circuit breakers, fallbacks), added monitoring before launch, wrote runbooks, and handled an incident gracefully. Mention the specific reliability metric you tracked (error rate, p99 latency) and how you improved it.

Tip: Pessimistic: acquire lock before read (SELECT FOR UPDATE) — prevents conflicts, but reduces concurrency. Optimistic: read without lock, check version on write (UPDATE WHERE version = X) — high concurrency, but requires retry on conflict. Razorpay uses optimistic locking for payment state transitions to avoid deadlocks at scale.

Tip: PCI DSS (Payment Card Industry Data Security Standard): 12 requirements for companies handling cardholder data. Key rules: never store CVV, encrypt PAN at rest/transit, quarterly vulnerability scans, segmented cardholder data environment. Razorpay is PCI DSS Level 1 certified — interviewers test basic compliance awareness.

Tip: Systematic approach: (1) Check distributed traces for the failed payment_id (Jaeger/Zipkin). (2) Check logs for the gateway response code. (3) Check bank API status page. (4) Reproduce with the same gateway + bank combination in staging. (5) If gateway-side issue, alert the partner and trigger failover to alternate gateway.