Pipeline

A pipeline is the automated sequence of stages a code change passes through on its way from a developer's commit to production. Modern CI/CD pipelines codify build, test, security scan, package, and deploy steps so every change moves through the same checks in the same order.

Typical stages

1. Source. Triggered by a commit, tag, or pull request from the version control system.
2. Build. Compile, transpile, bundle, or otherwise produce artifacts.
3. Test. Unit tests, integration tests, contract tests; ideally parallelised.
4. Static analysis. Linters, type checkers, SAST, license scanning, secret detection.
5. Package. Container images, language packages, or platform-native artifacts pushed to a registry.
6. Deploy. Apply to staging, run end-to-end tests, then promote to production.
7. Post-deploy. Smoke tests, synthetic checks, observability hooks.

Design considerations

• Parallelism. Run independent stages in parallel to shrink wall-clock time.
• Caching. Cache dependencies, layers, and test artifacts between runs.
• Selective execution. Run only what changed (monorepo affected detection).
• Approvals. Manual gates before production deploys for regulated or high-impact services.
• Reproducibility. Pipelines as code (YAML in the repo) so the same change yields the same run.

Common pipeline platforms

• GitHub Actions, GitLab CI, CircleCI, Buildkite, Jenkins, Drone, Tekton, Argo Workflows, AWS CodePipeline, Google Cloud Build