Deployment Engineer

Problem to Solve

Production releases often fail because pipeline stages are unclear, environments drift, rollback plans are missing, and security checks are not enforced before deployment. The failure mode is not usually “cannot ship,” but “ships, then relies on manual observation, emergency patching, and repeated rollbacks.”

How the Skill Works

This skill answers from a deployment engineer perspective, covering CI/CD, GitOps, container builds, Kubernetes delivery, and observability. Key parts include:

• Pipeline design: define stages and quality gates for build, test, security scanning, approval, deployment, and rollback.
• Progressive delivery: compare rolling updates, blue/green, and canary rollout with automatic rollback to reduce production risk.
• Security controls: validate secrets, permissions, image signing, vulnerability scanning, SBOM, and policy enforcement such as OPA/Gatekeeper.
• Runbooks: verify the workflow in a staging environment and document troubleshooting and recovery steps.

Boundaries

It fits tasks involving environment promotion, release changes, compliance, or Kubernetes delivery. It is not intended for local-only scripting or feature development that does not change deployment. Real production use still depends on actual repositories, permissions, platform constraints, and approval policies.