Navigate Ways of Working
← Index

Guides

Cheatsheets

Cheatsheet

Engineering Process Cheatsheet

The engineering lifecycle at a glance — development pipeline, sprint cadence, deployment flow, PR and feature flag policy, QA, rollout and rollback, delivery metrics, and AI applications.

Engineering Process

From refined item to production. Trunk-based, CI/CD, feature flags, AI-augmented.

Development Pipeline

Refinement
Squad
Problem, approach, acceptance criteria, size
Tech Design
Dev → Lead Dev
Implementation plan. 15–30 min conversation.
Build
Dev
Small commits to main behind feature flags
CI / CD
Automated
Tests, lint, SAST, security, AI review → deploy
Release Decision
PM + Lead Dev + QA
Async check. PM flips the flag.
Post-Merge Review
Lead Dev / Staff Eng
Learning & consistency. Not a gate.

Sprint Cadence (2 weeks)

WhenWhatDuration
Mon (start)Sprint planning30–45 min
Mid-weekRefinement (next sprint)45–60 min
ThuMid-sprint check15 min
Fri (end)Demo30 min
Every 4 wksRetrospective45 min
MonthlyArchitecture review60 min

Daily standups not mandated. Async updates. Squads choose.

Deployment Pipeline

D
Dev commits to main
↓
CI
Tests + lint + SAST + security + AI review
↓
CI
Auto-deploy to staging
↓
CI
Smoke tests
↓
CI
Promote to production

Target: commit → production < 30 minutes

Zero human intervention on happy path. Fully automated.

PRs & Feature Flags

Default: no PRs. Code goes to main via CI gates. Post-merge review for learning.

PR required for: shared platform APIs, database migrations, security-sensitive changes, new joiners (first weeks). 1 reviewer, 4-hr SLA, auto-merge if no response.

Flag TypeOwnerPurpose
EngineeringDevHide incomplete work
ReleasePMControl visibility
OperationalEngKill switch (no PM approval)

Clean up flags within 1 sprint of full rollout or kill. >60 days = escalate.

QA Across the Lifecycle

Refinement
Testability questions, edge cases, acceptance criteria
During build
Automated tests in parallel with dev, pairing on testability
At merge
CI quality gates. Exploratory testing for complex/risky changes
At release
Confirms quality for release decision
Post-release
Production monitoring, anomaly detection, regression detection

Rollout & Rollback

RolloutApproach
Non-trivial5–10% → 25% → 50% → 100% (monitor 24–48 hrs each step)
TargetedBeta users, specific segments, internal first
TrivialFull rollout immediately
RollbackSpeedWhen
Flag toggleSecondsFirst line of defence
Code revert~30 minWhen flag toggle insufficient

Rollback criteria defined before release, not during incident.

Delivery Metrics (DORA)

MetricSourceTarget
ThroughputProject trackerTrend over 4–6 sprints
Cycle timeProject trackerWithin 1 sprint
Deploy frequencyCI/CD logsDaily+
Change failure rateDeploy toolingTrending down
MTTRAlerting toolMinutes (flag toggle)

All derived from tooling. No manual entry. Never measure individual velocity.

AI in Engineering

StageAI Application
Tech designDev pressure-tests approach before Lead Dev conversation
DevelopmentCode generation, test writing, documentation
CI/CDAutomated code review, security patterns
QATest generation, anomaly detection, pre/post comparison
IncidentsLog analysis, timeline drafting, pattern detection
ReleasesRollout monitoring, auto-revert, release notes, stale flags
MetricsDashboards, weekly summaries, anomaly alerts

For full details, see the Engineering Process and DevOps Principles.