AI-driven optimization of CI/CD pipelines
summary
AI augments CI/CD by auto-generating pipeline models, scanning console output, triaging failures and proposing fixes that can open PRs and sometimes apply them, cutting manual investigation across 300 microservices and roughly 1,000 pipelines. Implementation uses GitHub Actions plus an LLM-driven scanner, security tooling and cost-monitoring to surface wasted resources and vulnerable code. Outcome: faster triage, frequent auto-suggested fixes (claimed ~90% useful), lower tool/runtime cost, but requires careful prompts, secret-handling and human review—never fully autonomous.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:37CI/CD pipeline challenges
- 5:50Pre-AI optimization efforts
- 8:23AI integration benefits
- 11:39AI-driven CI/CD demo
- 18:29AI limitations and risks
- 20:37Q&A
related talks
35mBeyond the chatbot, beyond the IDE, where can AI help your CI/CD
12mThe Synergistic Impact of Artificial Intelligence on DevOps
28mAI for Everything After Code
40mSecure by Default: Building an AI-Augmented, OSS-Powered Reusable CI/CD Pipeline
26mYour pipeline is the bottleneck
29mOpen-Source Framework for AI-Powered CI/CD Pipelines
24mUnlocking CI Performance in the Age of Agentic Development
19mRethinking CI & CD for Agentic Workflows