Letting AI Steal My Job - I'm Not Even Mad, This Is Amazing
summary
Built an automated code-review pipeline that runs on every PR using GitHub Actions, Python and LLM SDKs, calling Vertex AI / Gemini and GPT to generate review comments, unit tests and OpenAPI (Swagger) docs. It initially emits seven comments but is tuned to return top 3 comments (team later asks for 1–3), and generates tests via the IDE integration. The nightly job scans the repo, creates PRs with missing API docs, and the system works well though it needs iterative prompts and feedback.
AI-written summary · about
chapters (8)
- 0:00Introduction and speaker background
- 1:00AI tools revolutionizing development
- 5:58Understanding large language models
- 8:33Solving code review automation
- 12:55Building the PR review script
- 17:00Improving AI-generated code reviews
- 18:48Future AI integration plans
- 20:37Encouragement to start using AI
related talks
22mAI-Powered Code Reviews with Node and Vercel AI SDK
19mThe AI Code Review Bottleneck: Escaping the Velocity Trap
11mFrom Prompt to PR: Vibe Coding Meets AI Code Review
37mAI-Driven Developer Productivity at Scale
07mLightning Talk: Shift Left With AI: Smarter Code Reviews for DevOps...
26mAI-powered Code Review
41mImproving DevOps workflows with GenAI
09mImproving code reviews with AI agents