lancedb/vectordb-recipes
FreeExamples and tutorials for GenAI apps using LanceDB
FreeFree tier
Inputs: text, image, videoOutputs: text
About lancedb/vectordb-recipes
VectorDB-recipes is a GitHub repository by LanceDB that provides a comprehensive collection of examples, applications, starter code, and tutorials for building generative AI applications. It covers multimodal AI, retrieval-augmented generation (RAG), vector search, chatbots, AI agents, recommender systems, and more, all built using LanceDB—a free, open-source, serverless vector database that integrates seamlessly with Python data ecosystems (pandas, Arrow, Pydantic) and offers a native TypeScript SDK. The repository is divided into 'Examples' for quick prototyping and 'Applications' for ready-to-use apps, with interactive notebooks and scripts for hands-on learning.
Key Features
Step-by-step guides to build AI applications from scratch (e.g., RAG, agents)
Multimodal search examples across text, images, and video (CLIP, V-JEPA)
Multiple RAG implementations with various LLMs (Deepseek, OpenAI, Llama)
Vector search tutorials for efficient document retrieval
Chatbot creation with retrieval-augmented generation
Evaluation notebooks for measuring AI answer quality
AI agents with multi-agent collaboration (e.g., Fintech AI Agent)
Recommender system development using vector search
Concept tutorials explaining key AI techniques
Built on LanceDB: free, serverless, no setup required, integrates with Python and TypeScript ecosystems
Pros & Cons
Pros
- Free and open-source with no setup or server management required
- Integrates directly into existing Python data pipelines (pandas, Arrow)
- Native TypeScript SDK for serverless functions
- Wide variety of examples covering multiple AI domains (RAG, multimodal, agents)
- Interactive notebooks and scripts for hands-on learning
- Active community support via Discord and Twitter
Cons
- Primarily focused on the LanceDB ecosystem; examples may not be directly portable to other vector databases
- Some advanced examples require external API keys (e.g., OpenAI, Voyage AI) that involve costs
- Repository is a collection of recipes, not a production-ready framework; additional engineering may be needed for deployment
- Limited documentation beyond the README and example notebooks
Best For
Building a multimodal search engine for images and videosImplementing RAG on PDFs and documents with hybrid searchCreating a local RAG chatbot using Llama3Developing a fintech AI agent for financial queriesConstructing a recommender system for personalized suggestionsEvaluating and improving the accuracy of AI-generated answersExploring multi-head RAG architectures from scratch
FAQ
What is vectordb-recipes?
A GitHub repository by LanceDB containing examples, applications, starter code, and tutorials for building generative AI applications using LanceDB, a free open-source serverless vector database.
What types of examples are included?
Examples cover build from scratch, multimodal, RAG, vector search, chatbot, evaluation, AI agents, recommender systems, and concepts. Each includes interactive notebooks and scripts.
Is LancDB free to use?
Yes, LanceDB is free, open-source, serverless, and requires no setup. The repository itself is also free and open-source.
Do I need a GPU to run the examples?
Most examples run on CPU, but some may leverage LLMs that benefit from GPU acceleration or cloud APIs. The repository does not specify hardware requirements.
How do I get started with vectordb-recipes?
Clone the repository from GitHub and follow the instructions in the README. Each example folder contains its own notebook or script with setup guidance.