LLM Learning Hub

workspace/llm-course/home

Vector Database

A vector database is a system that stores embedding vectors with metadata and supports fast approximate nearest-neighbour search, forming the retrieval backbone of a RAG system.

What is Vector Database?

A vector database is a system that stores embedding vectors with metadata and supports fast approximate nearest-neighbour search, forming the retrieval backbone of a RAG system.

A vector database is a system that stores embedding vectors with metadata and supports fast approximate nearest-neighbour search, forming the retrieval backbone of a RAG system.

Where is it used?

Pinecone, Weaviate, Milvus, Qdrant, and pgvector are used in production RAG; FAISS and Chroma are popular for local prototyping with LangChain.

How to build it

Use `chromadb.Client().create_collection("docs")`, call `.add(ids=..., embeddings=..., documents=...)`, then `.query(query_embeddings=..., n_results=5)` to retrieve top chunks.