Appearance
DevOps / SRE perspective
Lens: Safe, observable paths to production — pipelines, deploy gates, telemetry, and incident response.
Phase by phase
| Phase | Your job | Key artifacts | Guides & SOPs |
|---|---|---|---|
| Plan | Flag infra/ops cost; tier validation | Tier in ticket | Runtime AWS |
| Define | Review IaC impact in ADRs | ADR with ops trade-offs | SOP-002 |
| Build | Maintain CI; ephemeral PR envs | Pipeline config (in your repo) | CI/CD · Identity & secrets · Linters |
| Verify | Ensure scan gates enforced | CI policy | SOP-005 |
| Release | Deploy; metric gate; rollback | Change record | SOP-006 |
| Operate | On-call; SLOs; synthetics | Dashboards, alerts | Monitoring · Dashboards · SOP-007 |
| Learn | Postmortem; alert tuning; baselines | Action items closed | SOP-008 · Monitoring as QA |
Accountabilities
- T1 prod deploy approval (with ARCH)
- Incident commander for Sev-1/2
- SLO definitions and error budgets
- Pipeline health and deploy freeze on budget burn
Deep dive: Operations & observability · CI/CD topic
Who you collaborate with
| Role | When |
|---|---|
| Developer | Failed pipelines, hotfix deploys |
| Architect | T1 deploy, cross-service blast radius |
| PO | Staging green but wrong product — escalate to PO not QA scripts |
| Security | GuardDuty, policy exceptions |
| Program manager | Cross-team release trains |
Pitfalls (DevOps view)
| Pitfall | Mitigation |
|---|---|
| Deploy without metric gate | Auto-rollback on canary |
| Alert fatigue | SLO-based paging only for T1 |
| Staging ≠ prod parity | Same container, scaled infra |
| Monitoring without runbooks | Every T1 alert links runbook |