: Workforce Efficiency with AI-Driven Vector Search
summary
Pipeline uploads resumes via API/Blob, uses Azure Document Intelligence to extract JSON, performs semantic chunking, calls an embedding model and stores vectors in Azure Cosmos DB for MongoDB vCore with a prebuilt vector index, then runs an IVF similarity search to surface matches and posts top chunks to GPT‑3.5‑turbo for final ranking. The production prototype reduces token consumption by about 1,986 tokens, returns correct bench candidates instead of stale first‑match results, and cuts LLM latency.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:24Why a Vector Database
- 2:46Native Vector Search
- 4:50Target Audience
- 12:10Semantic Chunking
- 14:17Vector Search
- 15:48Summary
related talks
46mAdvanced RAG with Azure AI Search - top questions answered
50mBuilding AI ready applications
01h 03mDrop the Bass with Embedding and Vectors in Azure AI Search
22mGenerative AI application stack and providing long term memory to LLMs
33mUsing the Amazing World
17mAzure Cosmos DB in AI Workflows: From Knowledge Retrieval to Microsoft Fabric Integrations
48mThe power of AI and Copilot for Azure Databases
01h 03mDrop the Bass with Embedding and Vectors in Azure AI Search