Workshop
Advanced RAG: Build AI That Knows Your Data
Your AI knows nothing about your data. Let's fix that.
About this workshop
Move past naive RAG and build retrieval systems that actually work. Covers hybrid search, reranking, chunking strategies, metadata filtering, and agentic RAG architectures using LlamaIndex and pgvector.
What you will learn
- Implement hybrid BM25 + dense vector search and understand when to use each
- Apply cross-encoder reranking to dramatically improve retrieval precision
- Design chunking and metadata strategies for long documents and mixed content types
- Build an agentic RAG pipeline that retrieves, reasons, and self-corrects
Who this is for
- Engineers who have built a basic RAG pipeline and know it isn't good enough
- AI teams building internal knowledge tools over proprietary document sets
- Developers who want to understand retrieval architecture at a production level
By the end
Before
RAG pipelines that retrieve the wrong chunks and hallucinate
After
Hybrid search with reranking that surfaces the right context every time
Before
Treating retrieval as a black box you can't debug
After
Full control over chunking, metadata, and search strategy
Before
A prototype that works with 100 documents
After
An architecture that scales to millions in production
About Yuki
Yuki Tanaka
AI Architect & RAG Specialist
Vetted by Maram
Yuki is an AI architect who has designed retrieval systems for enterprise clients across finance, healthcare, and legal services. A frequent contributor to LlamaIndex, he runs the RAG Patterns open-source repository with over 12,000 GitHub stars.
View full profile →What learners say
Reviews appear here once 3 learners have completed this session.