Presentation: Retrieval Augmented Generation (RAG) Pipelines
Abstract: Large language models provide powerful capabilities for generating natural language responses, but systems built solely on generative models often suffer from hallucinations, outdated knowledge, and limited domain accuracy. Retrieval Augmented Generation (RAG) addresses these challenges by combining modern information retrieval techniques with generative AI models. In a RAG system, relevant documents are retrieved from external knowledge sources at query time and provided to the model as contextual input, enabling responses grounded in verifiable information. This talk explains the architecture of RAG pipelines and walks through the stages involved in building them, including data collection, document processing, embedding generation, vector search, retrieval, and prompt construction. Implementation approaches and tooling in both the .NET and Python ecosystems will be discussed, along with considerations for data curation, web scraping, and responsible engineering practices when building AI systems. Attendees gain practical skills to build accurate, production-ready AI systems using Retrieval Augmented Generation and real-world tools.
Speaker(s): Scott Swindell
Virtual: https://events.vtools.ieee.org/m/558489