How we built an AI-powered code reviewer in 80 hours
summary
Prototype runs in a few hours; the backend takes 80 hours to stabilize using AWS API Gateway, Lambda, EventBridge, DynamoDB and Amazon Bedrock with Anthropic fallback, Clerk for auth and Stripe for payment. Real numbers: single LM calls take 10–30s, Claude 3.5 rate limit default 50/min (raised to 1,000), Lambda max 15m and many timeouts. Engineering fixes: analyze one file at a time and diff-only analysis, idempotent checkpoints, and avoid large-context failures around ~32KB.
AI-written summary · about
chapters (7)
- 0:00AI code reviewer intro
- 3:13Architecture and AWS setup
- 8:46Security and large PRs challenges
- 13:02Lessons learned
- 21:30Business model challenges
- 33:41Key takeaways and future
- 35:27Q&A
related talks
09mThe AI trick that cut AWS's issue triage time by 90% and caught regressions in 2 minutes
37mAI-Driven Developer Productivity at Scale
26mAI-powered Code Review
13mDeveloping with AI: Balancing Speed and Code Quality
45mThe AI Revolution in Software Development: From Code Writing to Quality Assurance
14mAI-Powered Code Review and Quality Assurance
40mCode Rot at Machine Speed: Preventing AI-Induced Technical Debt
46mThe amplified developer: zero to shipped in 30 mins with AI Coding Ass