The RAG workshop is a hands-on workshop for developers building AI over company documents. We go through the whole process from splitting documents to measuring quality, and I show what worked in my projects and what didn't.

Who it is for

For developers and tech leads who know Python or TypeScript and want to build a reliable RAG.

What participants take away

  • Hybrid search (vectors + BM25) and when to add a reranker
  • Citations, and declining to answer when the documents don't have the answer
  • Evaluation on a golden question set and regression tests in CI
  • Protection against prompt injection in documents

Agenda

  1. Chunking documents and contextual retrieval
  2. Hybrid search, reranker and knowledge graph
  3. Evaluation: MRR, nDCG and answer accuracy
  4. Security and deployment

Format

Half a day (4 hours) or a full day, online or on site at your company, up to 15 people. In Slovak or English.

Half day €1,400, full day €2,200 per group of up to 15 people. I am not VAT-registered, so prices are final.

Trainer

Adam Krampl, AI engineer with more than 8 years of experience. The material comes from projects I build and from my public ai-ml-learning repository. About

AI Act Art. 4 – AI literacy

Since 2 February 2025, companies that use AI must ensure their staff have sufficient AI literacy (EU AI Act, Article 4). A training with a certificate of completion is a simple way to document it.

FAQ

What results can you get?

In Construction Graph RAG, contextual retrieval and a reranker raised retrieval accuracy (MRR) from 76.1 to 96.5 % on 128 questions. At the workshop I show how to measure it on your side.

Do you use LangChain?

I show both approaches. In the ai-ml-learning repository I compare RAG in LangChain with one built from scratch, so you know when a framework helps and when it gets in the way.

Can we work on our own code?

Yes, I can build the workshop around your documents or an existing prototype.

Related