RAG: How Hard Could It Be ?
I've got some time to myself this week and there's bandwidth to venture into new areas. My friends and I were talking the other day, and they told me that Retrieval-Augmented Generation (RAG) is "fun" and worth exploring. Famous last words, but here we are.
But why write about it? Well, partially because it helps me organize my thoughts, partially because it serves as good notes for my future self and partially because it's therapeutic (we'll see about that).
So here it goes. It's all about rolling up my sleeves with RAG in a fun and practical manner. The idea is to learn through creating something concrete, trying every step of the RAG pipeline, and having fun in the process — or at least that's the plan.
The Mission
The objective of this project is pure academics, yet:
- Not merely reading about how RAG works, but actually getting it to work with my own files
- Investigating each component—document parsing, chunking, embedding, and smart retrieval — in a low-stress, "try-and-see" setting
- Building confidence to eventually use or modify RAG in larger, more complicated projects
But above all:
Create a safe playground to experiment with RAG in the real world and learn about its building blocks in the most friendly manner possible! No pressure, no "big" regulations — this is simply a RAG playground, friendly edition.
What's Next
I've lined up some online resources and I'm going to spend today going through them as pre-study. The real fun begins tomorrow when theory meets reality.
Stay tuned for the adventure!