The Challenge
CloudNova's engineering team of 65 was bottlenecked by a single Jenkins controller running 12 plugins with conflicting dependencies. Build queues routinely exceeded 40 minutes, and flaky integration tests caused a 30% retry rate.
Our Approach
- Pipeline Design — Created reusable composite GitHub Actions workflows for build, test, scan, and deploy stages.
- Ephemeral Runners — Self-hosted runners on spot instances with auto-scaling groups — zero queue wait, 70% cheaper.
- Security Shift-Left — Integrated Trivy for container scanning, SonarQube for SAST, and dependency-check in every PR.
- Artifact Management — Centralised Docker images and npm packages in AWS CodeArtifact with retention policies.
Results
Average build time dropped from 38 minutes to 6 minutes. The flaky-test detection system cut retry rates to under 3%. Engineers reclaimed 200+ hours per month previously lost to CI friction.
"Our developers actually enjoy the CI process now. PRs merge in minutes, not hours." — CTO, CloudNova