This workshop centers on the design and optimization of Retrieval-Augmented Generation (RAG) systems for building responsive, domain-specific search and question-answering applications. Participants will explore how to improve retrieval pipelines through techniques such as reranking, hypothetical document expansion (HyDE), chunking strategies with overlap, vector representation tuning, and small-to-big retrieval methods.
The session will highlight current trends in RAG development and provide practical guidance on how to adapt these systems to different types of data and user needs. Using a hands-on framework deployed on Jetstream infrastructure, attendees will scrape structured content from PDFs and build their own searchable applications—gaining experience with indexing, retrieval, evaluation, and deployment of high-performance, context-aware systems.
