RAG in the Era of LLMs

I chose RAG because it was the topic that interested me most at the time: I was reading its literature for my final-year project.

The workshop started from LLMs and their limits, hallucination in particular, then introduced RAG, went beyond the basic pipeline, and closed on industry applications, evaluation and open limitations. I ran a prepared hands-on part live to show each concept at work, and drew on Excalidraw to answer questions as they came. The idea attendees left with: knowledge is embedded, and the same embedding model embeds the question, so retrieval returns the passages closest to it.