Rethinking DevOps and DevSecOps with Generative AI: Insights
summary
Claims that generative AI reshapes DevOps by treating LLMs as services needing orchestration with LangChain and LangSmith, using RAG and vector databases to control hallucinations. Shows a hackathon that vectorized months of shell history to cluster commands and surface an Oracle dump — it works and reveals misuse patterns. Uses GPT‑4 embeddings (1536‑dim vectors), measures baseline correctness ~93% and hallucination ~4%, and monitors correctness, relevance, bias and ndcg as observability signals.
AI-written summary · about
chapters (9)
- 0:00Introduction and personal journey
- 2:17Why DevOps needs Generative AI now
- 6:09History of neural networks and AI
- 13:07Generic AI stack for organizations
- 18:09Observability in LLM systems
- 23:30Vector databases and embeddings
- 28:06Data preparation and chunking
- 29:53Workshop resources and demos
- 30:59Closing and Hugging Face demo
related talks
33mOperationalizing GenAI Applications with DevOps Practices
28mAIOPS the power of DevOps at scale
30mWhen AI Takes Over: DevOps Nightmare Unleashed
34mBeyond CodeGen: How AI & Machine Learning Are Transforming DevOps
41mEfficient DevSecOps workflows with a little help from AI
41mGetting real DevOps ROI with your Generative AI
38mGenerative DevOps
35mDynamic, Safe Ops Using AI to Generate Trusted Operations from Text Prompts