Demystifying the Data Stack of the Largest and Fastest Growing Payment Gateway in India
summary
Razorpay handles ~$150B TPV across 5M+ businesses while maintaining 99.99% uptime by using Aurora multi‑AZ with proxy servers, migrating a PHP monolith to Go microservices, and automating spot/on‑demand instance switching. Network defenses mitigate 542 DDoS events last year via granular load‑balancer and API rate limits plus merchant‑level time‑series AI. Fraud control combines a 400‑rule engine and risk models over 500 derived features; an RTO model cuts returns 50–70% and Razorpay even pays return shipping when the model errs.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:10Razorpay’s money movement
- 5:07Growth metrics and scalability
- 6:3499% platform availability
- 8:24Mitigating 542 DDoS attacks
- 10:50Payment fraud with AI and rules
- 13:00Reducing return-to-origin costs
- 15:54Revenue versus RTO balance
- 16:15Cost cutting at 60% scale
- 18:24Future roadmap and expansion
related talks
39mThe Science Behind Delightful User Experience
25mScaling AI for high-stakes, real-time payments
18mHow Microservices and a Key-Value Store Transformed India into a Cashless Society
33mFrom 0 to €500 Billion and Beyond, How Adyen Builds for Growth
47mKey Trends in the Shifting Threats to Payments as Blended Commerce Returns
15mGrowing
24mBuilding Planet-Scale Payment Platforms
13mAI-Powered Payment Systems