Register your interest

Save 100 on your pass

Call to action
Your text goes here. Insert your content, thoughts, or information in this space.
Button

Back to speakers

Jörn
Malich
Domain Engineering Lead
Schwarz digits
Jörn Malich, Domain Engineering Lead for GenAI & NLP, is a pivotal figure in Generative AI within the Schwarz Group. With a strong background in theoretical physics and a deep understanding of LLMs, he spearheads the practical application of GenAI, focusing on central technologies like RAG, Agents, Model Finetuning & Prompt Engineering. He also ensures the provision of sovereign GenAI models on the STACKIT platform for the whole Schwarz group, delivering secure and scalable enterprise solutions. At the upcoming GenAI Summit, he'll share insights on how Uncertainty-Aware LLM Agents can significantly enhance AI agents reliability and robustness in real-world scenarios.
Button
11 September 2025 14:20 - 14:50
Calibrating Trust: Engineering reliable genAI agents with uncertainty-aware LLM architectures
Agent systems are the new state of the art in GenAI, yet their real-world application is often hindered by unreliability due to LLM overconfidence and hallucinations. This talk introduces uncertainty quantification (UQ) as a critical method for estimating an LLM's intrinsic uncertainty. We'll present practical UQ methods and demonstrate how uncertainty-aware LLM pipelines can lead to more robust and reliable GenAI products. Through examples in Retrieval-Augmented Generation (RAG) and agentic pipelines, discover how UQ can help navigate LLM uncertainty and build trustworthy AI agents.