vectordb-recipes

Why we included this project

Working examples, starter code, and notebooks make this a practical on-ramp for teams getting hands-on with LanceDB and LLM applications. The recipes are grouped by the problems you actually build for: retrieval-augmented generation, multimodal search across text and images, conversational chatbots, and multi-agent systems, and many of them run directly in your browser on Colab. Because LanceDB needs no local setup and plugs into pandas and Arrow, the examples stay short enough to go from an idea to a working proof of concept in a single sitting. Even if you end up using a different vector store, the tutorials lay out the underlying retrieval and embedding mechanics clearly enough to adapt, including where reranking fits. The collection works both as a way to learn how RAG.