LlamaIndex RAG Pipelines

Build retrieval-augmented generation pipelines with LlamaIndex. Generates document loaders, index configurations, query engines, agent tools, and evaluation harnesses for production RAG systems.

This skill helps you build RAG systems with LlamaIndex. It generates document ingestion pipelines with chunking strategies, creates vector and knowledge graph indexes, builds query engines with hybrid search, implements agentic RAG with tool calling, configures reranking with Cohere or cross-encoders, and sets up evaluation with RAGAS metrics. Covers LlamaParse for complex document extraction and LlamaCloud managed pipelines.

When to use

Use when building RAG applications, ingesting documents into vector stores, implementing hybrid search, creating agentic query pipelines, or evaluating retrieval quality.

Examples

Document Q&A

Build a Q&A system over PDF documents

Create a LlamaIndex pipeline that ingests PDF documents, chunks them with sentence-level splitting, indexes in ChromaDB, and answers questions with source citations

Agentic RAG

Build an agent with multiple data source tools

Create a LlamaIndex agent that has tools for querying a knowledge base, searching recent emails, and looking up Jira tickets — with a router that picks the right tool