RAG. You’ve heard about it. Your entire family has heard about it. However, most LLM based projects never make it to production. In this talk, I’ll teach you some tricks on how to make a successful RAG app - so you don’t have to make the same mistakes.

This talk is about deconstructing the myths and misconceptions around a good RAG workflow. I will give the attendees practical advice on how to make their retrieval workflow better. It will be filled with practical advice that attendees can go out an apply directly.

Spoiler? It’s all about good Python.

Rough Outline:

Intro Slide
RAG - what is it?
The Langchain 4 liner and what’s wrong with it
It’s all about the data and validation
Leverage Pydantic validation and structured data
Evaluate very early
Representative vs. Keyword based
Summarization tricks
Pulling everything together
Logging everything
Final thoughts and QA

Note: The structure might change slightly

RAG: Tricks from the trenches.

Duarte Carmo