Improving Developer Experience: How We Built a Cloud Native Dev Stack for 100s of Engineers
summary
Razorpay builds a cloud-native DevStack that lets local code sync into EKS clusters using devspace sync, composes per-developer microservice sets with helmfile service-fleet and Helm hooks, mocks AWS with localstack, and routes traffic via header-based routing plus OpenTracing. CI remains GitHub Actions/Spinnaker and infra is Terraform/Atlantis. The stack supports hundreds of engineers across ~100 services and 1,500 monthly deployments and cuts feature lead time from ~5 hours to 30 minutes and iteration time from ~20–30 minutes to ~2 minutes.
AI-written summary · about
chapters (14)
- 0:00Start
- 9:20Goals
- 9:44Key Decision Factors
- 11:53How do we overcome network limitations?
- 12:51How can we avoid container restarts?
- 13:53How do we orchestrate multiple services?
- 16:37Helmfile Workflow
- 17:28How do we handle routing?
- 19:12Routing Overview
- 30:23Preview URL
- 30:38Ephemeral Databases
- 31:40Cost... really looking into that?
- 32:20Prometheus Monitoring
- 32:31Solution Overview
related talks
32mImproving Dev Experience: How We Built a Cloud Native Dev Stack At Scale
53mStreamline Cloud-Native App Development Using CDEs
44mCloud Native Applications with Remote Development Environments
46mIn Search of the Perfect Cloud Native Developer Experience
20mThe New Stack: building software without limits
29mSuccessful Cloud Native Platform Engineering – Hype vs Reality
52mCloud Native Developer Experience in 2025
41mRediscovering DevOps' Heartbeat: A Journey with Cloud Development Environments