Part 6 · 1 chapters · ~8 min

Review at Scale: Owners, SLAs, Bots and AI Reviewers

CODEOWNERS and ownership, review load balancing, response-time SLAs, automated checks that remove human work (formatters, linters, type checks, security scanners, size labels), AI reviewers and their limits, and metrics that improve review without gaming it.

Scaling review across many teams

code

# .github/CODEOWNERS: sensitive paths need their owners
/src/ledger/            @bank/ledger-team
/src/auth/              @bank/identity @bank/security
/migrations/            @bank/db-reviewers
/infra/                 @bank/platform

practice	detail
review SLA	first response within one working day; tracked, not punished
load balancing	auto-assign reviewers round-robin within owning teams
bots	formatting, lint, types, dependency and secret scanning, PR size labels, missing-test warnings
AI reviewers	useful for a first pass (obvious bugs, missing error handling, inconsistencies); they miss business context and can be confidently wrong; a human still approves (AI-native course part 4)
metrics	time to first review, time to merge, PR size distribution; avoid counting comments or approvals per person