Use your SAP HANA Cloud stored data to build solutions on AWS
summary
Architecture shows generating embeddings with Amazon Bedrock (29 models from seven providers, including Titan multimodal embeddings), storing them in SAP HANA Cloud vector engine and running top‑K similarity queries (cosine or L2) via SQL for retrieval‑augmented generation. Demo uses multimodal embeddings (text+images), LangChain/CAP integration, k‑NN search and returns top‑10 results; HANA Cloud on AWS leverages Graviton. Retrieval augmentation fixes LLM hallucinations by adding exact document chunks (example: corrected claim about native vector support in Q1 2024).
AI-written summary · about
chapters (9)
- 0:00Introduction and overview
- 1:10Use cases for vector embeddings
- 8:04Generating vector embeddings with Amazon Bedrock
- 13:54SAP HANA Cloud overview
- 17:03SAP HANA Cloud multimodel engines
- 23:27Vector database usage patterns
- 27:21Retrieval augmented generation workflow
- 35:08Multimodal models and applications
- 39:05Summary and next steps
related talks
33mVector Databases for GenAI applications
20mExplore what’s possible with new AWS generative AI services
29mCustomizing genAI applications for your business
37mBuilding Generative AI Applications
16mEmpower AI: Building Smarter Solutions with Embeddings
26mBuilding generative AI–enriched applications with AWS & MongoDB Atlas
22mGenerative AI application stack and providing long term memory to LLMs
48mHow to Make Smart Architecture Decisions when Building Gen AI Apps