LLMs That Know Your Data: A Hands-on Intro to RAG, Tuning, and Tracing

Large language models (LLMs) continue to reshape how we interact with technology, enabling natural, conversational interfaces across industries. These models, trained on vast quantities of data, support a wide range of natural language processing (NLP) tasks—summarization, question answering, sentiment analysis, entity recognition, and more. But do you know how to apply these models to your business data—safely, securely, ethically, and profitably?

In this hands-on workshop, you’ll gain the practical foundations to “talk with your data” by implementing an LLM-powered information retrieval and question-answering system. You’ll explore key components such as model selection, the art of prompt tuning, and how to build a modern retrieval-augmented generation (RAG) pipeline. You’ll also learn techniques for improving system performance, from incorporating domain-specific data to choosing the right retrieval strategy—and critically, how to trace and observe your system’s behavior to ensure reliability, transparency, and safety at scale. Afterwards, you will be equipped to apply these skills in your own business context—with your own data and tools—to unlock real value from LLMs. While we’ll work in a Google Colab notebook during the session, the skills taught are broadly applicable across platforms and environments.