Principal-level platform design question — architecture, developer experience, and governance.

Source: GitHub + branch protection + CODEOWNERS + signed commits

CI: GitHub Actions reusable workflows (canonical pipeline per language)

Registry: GCP Artifact Registry + Container Analysis auto-scan

CD: ArgoCD + two-repo GitOps (app repo / manifests repo)

Secrets: External Secrets Operator + GCP Secret Manager

Security: Binary Authorization + Workload Identity

Preview: Ephemeral environments per PR (short-lived GKE namespace)

Metrics: DORA dashboard (deploy freq, lead time, CFR, MTTR)

Platform team owns: reusable workflow templates, manifests repo structure, ArgoCD ApplicationSet config, DORA metrics dashboard. Service teams only write application code — CI/CD is inherited by using the platform Dockerfile conventions and reusable workflows. Goal: a new service is production-deployable within 1 day of creation, using the same pipeline as all other services. Platform SLO: commit to production in under 30 minutes including manual approval.

Developer experience first: A fast, reliable pipeline is the platform team's product. Track pipeline P95 duration and treat improvements as feature work.