Razorpay processes ₹10+ lakh crore in payments annually for 8M+ businesses — making it one of India's most critical digital infrastructure companies. At a $7.5B valuation, engineers here solve hard problems: sub-second transaction latency, 99.99% uptime for payment rails, and real-time fraud detection at scale.
Company Overview
Founded in 2014 by Harshil Mathur and Shashank Kumar (IIT Roorkee alumni), Razorpay has evolved from a payment gateway to a full financial infrastructure platform. The product suite includes Razorpay Payment Gateway, Razorpay X (business banking), Razorpay Capital (lending), and Magic Checkout.
| Detail | Info |
|---|---|
| Headquarters | Bengaluru, Karnataka |
| Other India Offices | Mumbai, Delhi NCR, Hyderabad, Chennai |
| Employees | ~3,500+ (India-first, global offices in US, Malaysia) |
| Primary Tech Stack | Java (core), Go, Python, Node.js, React, MySQL, Redis, Kafka, Kubernetes, AWS |
| Interview Difficulty | Hard — comparable to Swiggy/Flipkart; harder than Freshworks/Zoho |
| Valuation | $7.5B (Series F, 2021) |
Hiring Tracks & Salary
| Role | Experience | Base | ESOP Grant | Total CTC |
|---|---|---|---|---|
| SDE I | 0–2 yrs | ₹14 – 20 LPA | ₹10 – 20 L (4-yr) | ₹18 – 28 LPA |
| SDE II | 2–5 yrs | ₹25 – 40 LPA | ₹25 – 50 L (4-yr) | ₹32 – 55 LPA |
| Senior SDE | 5–8 yrs | ₹40 – 60 LPA | ₹60 – 1.2 Cr (4-yr) | ₹55 – 80 LPA |
| Staff Engineer | 8–12 yrs | ₹60 – 90 LPA | ₹1 – 2 Cr (4-yr) | ₹80 – 110 LPA |
| Principal Engineer | 12+ yrs | ₹80 – 120 LPA | ₹2 Cr+ (4-yr) | ₹100 LPA+ |
Razorpay is not yet publicly listed. ESOPs are a key part of compensation but are illiquid. Razorpay has run two secondary sale opportunities for employees (2020, 2022), giving partial liquidity. Factor this into your decision — ESOPs are valuable but not guaranteed to convert to cash on a predictable timeline.
Interview Process
Online Assessment
HackerRank / Razorpay portal: 2–3 coding problems (Medium–Hard). 90 mins. Strictly proctored. Expect at least 1 Hard problem.
Technical – DSA
Live coding on CoderPad. 1–2 problems. Expect follow-ups: "Optimise it further", "What if the input is 1B items?". 45–60 min.
Technical – Concepts
CS fundamentals: concurrency, DB transactions, networking, caching. Past project deep-dive with "How did you scale it?" probing. 45–60 min.
System Design
Mandatory for 2+ yrs experience. Payment-domain design: transactions, idempotency, reconciliation. HLD + LLD expected. 45–60 min.
Hiring Manager / Culture
Behavioural: ownership, cross-team collaboration, past failures. Razorpay's culture focuses on "First Principles" thinking. 30–45 min.
HR / Offer
CTC discussion, ESOP explanation, joining date, team matching. Multiple competing offers strengthen your position significantly.
OA Patterns
Razorpay's OA is notably harder than Freshworks or Zoho. Expect at least one Hard-level problem.
| Topic | Frequency | Notes |
|---|---|---|
| Graphs (BFS/DFS/Dijkstra) | Very High | Shortest path, connected components, topological sort |
| Dynamic Programming | Very High | Interval DP, tree DP, bitmask DP for Hard problems |
| Trees (BST + General) | High | Segment trees for range queries in senior roles |
| Sliding Window / Two Pointers | High | Often the "easy" warm-up in a problem set |
| Heaps / Priority Queues | High | Top K, scheduling, median-of-stream problems |
| Binary Search on Answer | Medium–High | "Minimum maximum" type problems |
| Greedy Algorithms | Medium | Interval scheduling, activity selection |
| Tries | Medium | Autocomplete, longest prefix matching |
| Concurrency Problems | Medium (senior) | Producer-consumer, deadlock prevention |
Real Questions Asked at Razorpay
Given a weighted directed graph representing payment routes between banks, find the minimum cost to route a payment from source bank to destination bank. If no path exists, return -1. Constraints: up to 10,000 nodes. (Dijkstra's with negative weight awareness.)
Given a list of transaction amounts, find the maximum sum of a non-contiguous subsequence such that no two selected amounts are at adjacent indices, AND the total does not exceed a given fraud limit. (Variant of house robber with an upper bound constraint.)
Given a log of transactions with timestamps and amounts, find the time window of length W seconds that has the highest total transaction volume. Then: "What if transactions can be out of order by up to 5 seconds?" Redesign your solution.
Implement a thread-safe LRU cache in Java. Then: "Now make it distributed — multiple instances need to share the same cache state." Discuss synchronisation, consistency trade-offs, and whether you'd use Redis. (No coding required for distributed part — design discussion.)
"Walk me through what happens at the database level when 1,000 concurrent users try to book the last seat on a flight." Cover: optimistic vs pessimistic locking, SELECT FOR UPDATE, deadlock scenarios, and how you'd avoid them in Razorpay's payment processing context.
"Design Razorpay's payment gateway. A merchant calls your API with an order; you need to route it to the correct bank, handle failures, ensure exactly-once processing, and return the result within 3 seconds." Cover: API design, idempotency keys, retry logic, database schema, and monitoring.
"Design a daily reconciliation system that compares Razorpay's internal transaction records with bank settlement files. Handle 10M+ transactions/day, detect discrepancies, and flag them for human review." (Tests distributed data processing knowledge: batch jobs, checkpointing, alerting.)
"Tell me about a time you identified a problem that nobody else on your team noticed. How did you convince others it was real?" (Razorpay loves "first principles" thinkers who surface issues proactively.)
Fintech Domain Knowledge (Must-Know)
Razorpay interviewers often ask domain questions, especially in system design. You don't need to be a fintech expert, but these basics will set you apart:
Payment Fundamentals
- Payment gateway vs payment processor vs acquirer — three different roles in a transaction
- UPI architecture — VPA, NPCI, PSP (Payment Service Provider), UPI flow diagram
- Card network flow — issuer bank, acquirer bank, Visa/Mastercard switch, authorization vs capture
- Net banking — redirect flow, callback handling, timeout scenarios
Reliability & Correctness in Payments
- Idempotency — why every payment API needs idempotency keys (retries without double-charging)
- At-least-once vs exactly-once delivery — and how to achieve exactly-once in Kafka
- Two-phase commit — how Razorpay coordinates state between its DB and the bank
- Circuit breaker pattern — essential when bank APIs go down during peak load (Diwali sales)
- Saga pattern — distributed transaction management across payment, order, and notification services
Prepflix Success Story: Infosys → Razorpay
Rahul Verma, a Prepflix student, spent 3 years at Infosys on a banking domain project — giving him just enough fintech context to resonate in the Razorpay hiring manager round. After 4 months of DSA + system design prep, he cleared all 5 rounds in his first attempt. Rahul's advice: "I failed the OA the first time because I tried to solve the hard problem first and ran out of time. Second attempt: I solved the medium problem in 25 minutes, then attacked the hard one with the remaining 65 minutes. That order change made all the difference."
90-Day Roadmap: Service Company → Razorpay
Strong DSA Foundation
- Arrays, Strings, Two Pointers, Sliding Window
- Trees, Graphs (BFS, DFS, Dijkstra, Topo Sort)
- Heaps, Priority Queues
- 100 LeetCode Medium problems
- Read Razorpay Engineering Blog for domain context
Hard DSA + Java Concurrency
- DP patterns: interval, tree, bitmask (30 Hard problems)
- Java concurrency: locks, futures, thread-safety patterns
- DB: ACID, transactions, indexing, sharding strategies
- Fintech domain: read UPI spec, understand payment flows
- System design patterns: idempotency, circuit breaker, saga
System Design + Mock Rounds
- Design: Payment gateway, fraud detection, reconciliation
- Design: Rate limiter, notification service, URL shortener
- 5 mock interviews (at least 2 with payment domain focus)
- Prepare behavioural stories: first principles, ownership
- Apply via LinkedIn/referral; warm intro = 10× conversion