LlamaIndex
Data framework for connecting LLMs to custom data sources
View on GitHubWhat it solves
Turning heterogeneous data sources into LLM-queryable indices without writing custom parsers and retrieval pipelines for each format.
Overview
LlamaIndex provides abstractions for indexing, retrieving, and synthesizing information from custom data sources (PDFs, APIs, databases, web pages) into LLM-ready context. Core concepts: nodes (chunks), indices (retrieval structures), and query engines (retrieval-to-synthesis pipelines).
Key facts
- Language
- Python
- License
- MIT
- Maturity
- Community canon
- Maintainer
- run-llama
- Reviewed
- 2026-06-11
Where to start
The RAG guide explains the index → retriever → query engine pipeline that underpins almost every LlamaIndex application.
docs/understanding/rag.mdTechnologies
More in this domain
ChromaDB
Open-source embedding database for AI applications
Persisting, querying, and filtering vectors without standing up a full database stack.
pgvector
Vector similarity search extension for PostgreSQL
Infrastructure overhead for teams already on Postgres: semantic search without adding a separate vector database to operate.