CI/CD & Release Infrastructure

Deploy time: 45min → 8min. Six months without a release-caused incident.

Redesigned the entire release pipeline with GitHub Actions + Jenkins + Terraform. Parallelized test suites, added Docker layer caching, and moved to immutable infra. Deploy time: 45min → 8min. Production incidents from releases: 2/month → 0 for 6 consecutive months.

deploys took 45 minutes, required 3 people, and broke prod twice a month

Highlights

• Parallelized Jest test suites across 4 GitHub Actions workers using jest --shard; test step went from 18min to 4min.
• Docker BuildKit layer caching in CI: node_modules layer only rebuilds on package-lock.json change — image build dropped from 9min to 90sec on cache hits.
• Terraform workspaces for environment parity: dev/staging/prod are identical except for variable overrides — 'works on staging, breaks on prod' eliminated.
• Feature flags (LaunchDarkly) decoupled deploy from release: code ships dark, product turns it on — zero-risk rollout for high-risk features.
• CloudWatch + New Relic dashboards with alert routing to PagerDuty; P1 MTTD (mean time to detect) dropped from 40min to under 5min.

See full portfolio: https://portfolio.kunal-gautam-570.workers.dev