Building Performant RAG Applications for Production

Learn how to build production-ready Retrieval Augmented Generation (RAG) applications in this 32-minute conference talk by David Carlos Zachariae at GOTO Copenhagen 2024. Discover why RAG is essential for enhancing Large Language Models with specific knowledge outside their training data, and explore a step-by-step approach to developing performant RAG systems. Follow along with practical demonstrations through four iterations: starting with a simple implementation, progressing to handling multiple documentation categories, managing unstructured documentation, and finally implementing dynamic context and actions. The presentation includes live demos at each stage, showing how to overcome common challenges in transitioning RAG applications from prototyping to production. Gain valuable insights on creating flexible, reliable, predictable, and scalable RAG pipelines that can handle diverse and complex tasks in real-world scenarios.