Hands-on Practice
summary
Giza Systems describes a deployed RAG pipeline that uses vector database + retriever with keyword, semantic and metadata filtering to turn docs and logs into embeddings, feed an LLM, then validate outputs; teams report 90–95% precision on retrieval-driven answers. Practical builds include automated incident triggers, layered root‑cause analysis that flagged a two‑week ransomware intrusion, and onboarding helpers that cut manual triage — most projects run on offline fine‑tuned models (LoRA/RAFT/BAFT variants) to meet data‑governance constraints.
AI-written summary · about
chapters (11)
- 0:00Introduction to AI and DevOps
- 3:45Understanding the Chinese Room analogy
- 6:10Introducing the AI model analogy
- 9:27RAG technology overview
- 10:57Vector databases in RAG systems
- 13:10How retrievers work in RAG
- 17:15Fine-tuning in AI models
- 18:18Raft and BART fine-tuned models
- 20:07Code generation impact on burnout
- 21:35Automating operations with AI
- 23:46Demo and Q&A
related talks
33mOperationalizing GenAI Applications with DevOps Practices
12mDeploying AI Literature Agents at Scale
31mAIOps: Automating and Scaling ML and LLM Workloads in Production
27mGrounding AI in reality: Building reliable infrastructure development workflows
13mBeyond Manual Firefighting
38mData Secrets from a Platform Engineer
41mEfficient DevSecOps workflows with a little help from AI
29mHow to automate the deployment of ...