Weeks 1–2 (Observe & orient)
- Goals: Meet team, access accounts, read runbook, runbook sanity-check, shadow primary on-call.
- Deliverables: Access checklist, annotated runbook gaps, list of top-10 alerts & owners.
- Stakeholders: On-call SRE, service owner, Dev lead, infra admin.
- Metrics: Able to reproduce 2 common alerts in dev; runbook covers those.
Weeks 3–4 (Deep code + infra dive)
- Goals: Read critical service components (startup, config, health checks), CI/CD pipelines, deployment playbooks.
- Deliverables: Architecture diagram, dependency map, short doc: “how deployment works.”
- Stakeholders: Dev lead, CI/CD engineer.
- Metrics: Can walk through a deployment end-to-end and explain rollback.
Weeks 5–7 (Monitoring & SLOs)
- Goals: Audit alerts, reduce noise, map metrics to user impact, propose SLOs.
- Deliverables: Alert-priority matrix, 1 SLO proposal with error-budget policy, 2 alert tuning PRs.
- Stakeholders: Product owner, SRE manager, analytics.
- Metrics: Reduced false-positive rate on tuned alerts by target (e.g., 30%); SLO draft approved.
Weeks 8–10 (Automation project)
- Goals: Deliver small automation that saves on-call time (e.g., automated remediation or runbook-runner).
- Deliverables: Tested automation PR, CI tests, deployment plan, monitoring for automation.
- Stakeholders: Security, QA, on-call SREs.
- Metrics: Automation reduces time-to-resolve for targeted alert by ≥20% in staging.
Weeks 11–12 (Ownership & solo readiness)
- Goals: Take increasing on-call shifts as primary with mentor shadowing off-hours; run 1 post-incident review.
- Deliverables: On-call checklist updates, incident postmortem, knowledge-transfer session.
- Stakeholders: Mentor SRE, service owner, rotation team.
- Readiness metrics (must meet before solo release):
- Successfully handled 3 distinct incident types end-to-end with mentor signoff.
- Runbook coverage ≥90% of top incidents and proven playbook execution.
- Alert noise below agreed threshold and automation deployed and monitored.
- Peer/manager approval after checklist review.
Ongoing (post-90)
- Quarterly SLO review, retire / extend automation, incremental reliability projects.
This plan balances shadowing, learning code, improving monitoring, and delivering automation with measurable deliverables, stakeholders identified for approvals, and concrete readiness gates before solo on-call.
