Vector databases explained, featuring clowns, embeddings, neural networks, feature extraction, semantic search and Retrieval Augmented Generation (RAG).
I walk through what a vector database is, by first explaining the types of problems that vector databases solve, as well as how AI "thinks".
I use clowns as an example of a large corpus of training data from which we can extract high level features, and I discuss architectures such as semantic search and RAG.
Applied AI at WorkOS. Formerly Pinecone, Cloudflare, Gruntwork. Full-stack — databases, backends, middleware, frontends — with a long streak of infrastructure-as-code and cloud systems.
Discussion
Giscus