← Module 05: Engineering delivery and operations MODULE 05 · ENGINEERING DELIVERY AND OPERATIONS · 40 MINUTES

Engineering delivery and operations

Present slides Full screen, one slide at a time · 5 slides
THINK LIKE A CTO · MODULE 05 · DELIVERY 01

How a change reaches customers

  1. GITEvery change is recorded: who, what, when and why
  2. REVIEWAnother person reads it before it is merged
  3. AUTOMATED TESTSHundreds of checks run on every change, in minutes
  4. CI/CDThe pipeline builds, tests and deploys the same way every time
  5. STAGINGA safe copy of production to try it first
  6. PRODUCTIONReleased in small steps, watched, and reversible

Small, tested, reversible changes, many times a week, are safer than big releases once a quarter.

THINK LIKE A CTO · MODULE 05 · DELIVERY 02

Environments and release controls

  1. DEVELOPMENTEach developer’s own copy, with fake data
  2. STAGINGLike production, with no real personal data
  3. PRODUCTIONReal customers, real data, limited access
  4. RELEASE CONTROLSReviews, passing tests, feature flags and a rollback plan

Real customer data stays in production. Copying it to test with is a breach waiting to happen.

THINK LIKE A CTO · MODULE 05 · OPERATIONS 03

Observability and incidents

  1. LOGSWhat happened, in order, for investigating problems
  2. METRICSNumbers over time: speed, errors, bookings per minute
  3. ALERTSA person is told when something breaks, before customers notice
  4. INCIDENT RESPONSEWho leads, how customers are told, how it is fixed
  5. REVIEW WITHOUT BLAMEWhat we learned, and what changes so it can’t happen again

If nobody is told when production breaks, your customers will tell you, publicly.

THINK LIKE A CTO · MODULE 05 · ENGINEERING WORK 04

AI-written code: faster, not free

What AI coding tools change

  • Drafts, tests and fixes in minutes
  • Small teams can build much more
  • Non-engineers can prototype and explore

What a CTO still insists on

  • A person who understands and reviews every change
  • Tests, security checks and the same release controls
  • No secrets or customer data pasted into AI tools

Organise engineering around review, tests and ownership, whoever or whatever writes the code.