Engineering Playbook
Practical templates and checklists I use to bring governance and clarity to engineering work. Download, adapt and use them in your team — no attribution required.
Templates
design.md
Design document template: context, scope, requirements, proposed architecture, decisions, risks, test strategy, rollout and rollback plans. The single most effective tool for aligning humans — and AI assistants — before code is written.
Download templateADR — Architecture Decision Record
Lightweight record of one architectural decision: context, options considered, decision and consequences. Makes the "why" auditable long after everyone forgot the meeting.
Download templateTechnical RFC
Proposal template for significant changes: problem, proposal, alternatives, impact analysis, security and operational considerations, open questions. Built for async, written decision-making.
Download templatePRD — Product Requirements
Product management template: problem evidence, success metrics, jobs to be done, scoped requirements with acceptance criteria, launch and measurement plans. Aligns product, engineering and design before anything is built.
Download templateDigital Product Creation Flow
Stage-gated flow from idea to operation: opportunity framing, discovery, definition, build, launch and operate — each stage with a clear gate decision (persevere, pivot or stop). Rigor scales with the size of the bet.
Download templateARB — Architecture Review Board Playbook
Operating model for an ARB that serves teams instead of blocking them: scope, composition, 48h pre-reads, 45-minute review format, decision rules with SLA, ADR artifacts and the anti-patterns to refuse.
Download templateChecklists
Cloud cost (FinOps)
- ✓Cost visibility per workload/service — not just per account
- ✓Right-sizing review: CPU/memory utilization vs. allocation
- ✓Storage lifecycle policies and orphaned resources sweep
- ✓Reserved capacity / savings plans coverage for steady workloads
- ✓Architecture review: can a redesign eliminate the cost entirely?
- ✓Cost regression alerts wired into the team’s routine
Rollout & rollback
- ✓Feature flag or staged rollout path defined
- ✓Database migrations backward-compatible with previous release
- ✓Rollback procedure tested — not just documented
- ✓Observability: dashboards and alerts for the new path before traffic
- ✓Success metrics and abort thresholds agreed before deploy
- ✓Communication plan: who is notified, when, through which channel
Observability minimum
- ✓Golden signals (latency, traffic, errors, saturation) per service
- ✓Structured logs with correlation IDs end to end
- ✓Alerts tied to user impact, not raw thresholds
- ✓Dashboards reviewed in incident drills, not built during incidents
- ✓P99 — not averages — as the latency conversation
Financial-environment security
- ✓Segregation of duties: no single person deploys and approves money-moving changes
- ✓Every financial mutation behind an explicit state machine with audit trail
- ✓PII masked at rest and in logs; access on least-privilege with periodic review
- ✓Secrets in a vault, rotated; no credentials in code, config or CI logs
- ✓Idempotency keys on every payment operation — retries must be safe
- ✓Reconciliation jobs comparing internal ledger vs. provider statements
- ✓Pen test and dependency/secret scanning wired into CI, not annual rituals
- ✓Incident playbook for financial impact: freeze, assess, communicate, reconcile