AI Academy I Advanced Retrieval-Augmented Generation (RAG)

#Talks
Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by integrating information retrieval, overcoming key limitations such as hallucinations, outdated knowledge, and lack of access to private company data. This makes RAG a highly valuable business application across different industries. However, major challenges remain, including retrieval efficiency, source quality, and integration of structured and unstructured data. This AI Academy explores advanced RAG techniques that aim to solve these challenges. As always at the AI Academy, we invite a speaker from research (Christopher Tauchmann) and a speaker from industry (Marcel Rosiak) to shed light on an innovative AI topic from both sides.
Christopher will provide an overview of Retrieval-Augmented Generation (RAG), covering fundamental concepts and key advancements in retrieval, and domain specific fine-tuning. His talk will explore how these developments can enhance the relevance of generated content.
Marcel will present two real-world case studies of advanced RAG applications, focusing on (1) optimizing a small, German-language RAG system with targeted training methods and on (2) integrating knowledge graphs into the retrieval process. His talk will cover business strategies for successfully implementing these advancements in enterprise settings.

#Speakers
Christopher is a postdoctoral researcher at the AIML Lab at TU Darmstadt and hessian.AI. His work focuses on post-training LLMs, with an emphasis on reasoning.

Marcel leads AI innovation at embraceableAI with a focus on developing practical, enterprise-ready language model solutions. Currently he is heading GRAG project, a collaborative effort to advance German-language AI capabilities.