What to Expect
Building the data foundation for GenAI using VeloDB:
- Ingest data from open data lakes into VeloDB using WhaleOps, build a vector index in Zilliz, and then provide GenAI features based on Amazon Bedrock.
A high-level view of GenAI framework:
- Get metadata and history query data from VeloDB, embed the data, get vectors and save them in vector database. When users give a prompt, use the embedding framework to get a vector, do similarity search in the vector database, and then use the Bedrock Titan framework to help retrieve the data from VeloDB.
How VeloDB helps?
-
VeloDB provides a unified data management framework, integrating data silos and providing data security and governance features. For example, it can integrate data from Redshift, MySQL, PostgreSQL, S3 and unstructured data such as files and logs. Additionally, it is integrated with Apache Ranger to provide centralized administration, and Atlas to offer data governance capabilities.
-
VeloDB can deliver rapid batch processing and is blazing-fast in online data retrieval. It also provides various storage and computing optimizations for semi-structured data. It offers a UDF framework and supports remote UDF.
