Streamlining data governance
summary
Concrete recipes for governing GenAI using AWS: walks through a prompt catalog plus versioning (MLflow v2.21 or Bedrock prompt manager) and recommends SageMaker Unified Studio/notebooks for experimentation, plus an event-driven data fresher using EventBridge → SageMaker to keep knowledge bases current. Shows pipelines for chunking/embeddings, retrieval, generation and evaluation, prescribes AB/blue deployments with human review gates and CloudWatch alerts. No hard ROI numbers; advice is pragmatic and production-focused.
AI-written summary · about
chapters (14)
- 0:00Introduction to GenAI governance
- 1:28MLOps vs GenOps shift
- 3:39Governance in RAG and agent ops
- 5:45Prompt governance challenges
- 8:33Prompt governance solutions
- 12:30RAG knowledge governance
- 14:57RAG pipeline governance
- 18:49Agent governance complexity
- 20:39Agent composition and tools
- 22:01Agent evaluation frameworks
- 23:03Agent testing and release
- 24:47Best practices for GenAI governance
- 26:11Getting started with governance
- 27:36Closing and improvisation
related talks
01h 00mSecurely accelerating generative AI innovation
48mHow to Make Smart Architecture Decisions when Building Gen AI Apps
45mServerless & Event-driven Patterns for GenAI
23mGenAI-Driven Knowledge Bases: Optimizing AWS Support and Cloud Ops
33mBuild, train, and deploy ML models using Amazon SageMaker
21mSmarter, faster Data analytics with generative AI and Machine Learning
47mOutsmarting the Genie: Smart Architecture Decisions for GenAI Apps
18mNavigating the growth frontiers of generative AI