The Amazon Software Development Process
summary
Explains why Amazon breaks a monolith into small services, organizes two‑pizza teams with end‑to‑end ownership, and runs “millions of deployments” via CI/CD; describes the internal Apollo deployer, a one‑box pessimistic rollout that bakes changes per machine/availability zone/region to limit blast radius, and weekly ops and business-review rituals that surface incidents; shows GenAI in Amazon Q developer for code completion, security scans and a code‑transformation that upgraded ~1,000 Java apps in days.
AI-written summary · about
chapters (12)
- 0:00Introduction to Amazon culture
- 2:35Transition to microservices
- 6:18Two-pizza team philosophy
- 12:03Self-service developer tools
- 15:55Continuous delivery at scale
- 18:48Pessimistic deployment strategy
- 24:02Automated quality assurance
- 26:10AI in software development
- 30:55Optimizing the inner loop
- 33:25Operational excellence
- 38:22Business review mechanisms
- 39:35Working backwards process
related talks
43mThe Amazon Software Development Process
50m Drive change with Amazon’s culture of innovation
25mOur Secret to Innovation
41mTowards Operational Excellence
45mHow Amazon Innovates: Best Practices for your Organization
31mHow we leverage mechanisms at Amazon?
33mSoftware development practices for companies that want to make an impact
38mWorking backwards: Amazon’s approach to innovation