Build your own advanced AI solutions on Azure Databricks
summary
Demo shows a working RAG pipeline on Azure Databricks that ingests research PDFs into Unity Catalog, chunks and embeds text with BGE large, builds a vector search index (PDF_bold_index) and ties a retriever to DBRX Instruct via the Foundation Model API; the chain returns an accurate ARES definition where base models fail. Pipeline registers the chain with MLflow, exposes it through model serving with an 80/20 traffic split and inference tables for auditing.
AI-written summary · about
chapters (9)
- 0:00Building AI on Azure Databricks
- 4:11Why Azure Databricks for GenAI
- 10:01Designing generative AI applications
- 14:34Demo: AI playground & RAG setup
- 18:03Unity Catalog for governance
- 21:19Vector search index for RAG
- 24:45Register & serve RAG model
- 29:43Monitor AI pipelines
- 31:30DBRX & next steps
related talks
50mBuilding AI ready applications
49mBuilding Advance Analytics pipelines with Azure Databricks
23mBuilding AI-Driven Real-Time Lakehouse APIs
19mAI Driven Data Enrichment with Azure Cognitive Search
13mBuilding High-Performance AI Literature Processing
44mSupercharge your data with Azure AI Search and OpenAI
50mAzure AI, Power new possibilities for every organization
01h 58mBuild, Evaluate and Deploy a RAG-Based Retail Copilot with Azure AI