Free live bootcamp
Build Production-Grade RAG for Finance
A hands-on live class on chunking, hybrid retrieval, reranking, and grounded answers with citations.
Most RAG demos fall apart on real financial documents. Tables get shredded, retrieval returns the wrong quarter, and the model answers confidently with no source. In this live class we build a retrieval pipeline that survives all three, and we do it on real filings rather than toy text.
What we build in these two hours
- Why a normal RAG pipeline breaks on financial documents: 10-K filings, earnings call transcripts, and tables full of numbers
- Chunking strategies that do not destroy tables and financial context
- Hybrid retrieval: BM25 plus vector search with Qdrant, and when each one wins
- Reranking: how to go from roughly relevant to actually correct
- Grounding, citations, and hallucination guardrails, which are non-negotiable when the answer is a number
- A full live Q&A at the end
Who it is for
- Engineers who have built a RAG demo and watched it fail on real documents
- Data scientists moving from notebooks to something that has to hold up
- Anyone working with financial filings, reports, or long PDFs
What you need
- Comfortable writing Python
- Basic understanding of embeddings and vector search
- A Google account, since the class runs on Google Meet
Questions
Is it really free?
Yes. The class is free to attend. We build the system live and answer questions until they run out.
Will I get a certificate?
Everyone who attends the live class receives a Certificate of Participation by email.
What do I need installed?
Nothing before the class. The setup steps go out in the welcome email so you can follow along while we build.