Speakers

Josefa Stoisser is an AI Research Scientist at Novo Nordisk, where she architects large language models and autonomous agentic systems to accelerate drug discovery R&D. Holding mathematics degrees from the University of Cambridge and ETH Zurich, her work is dedicated to bridging the divide between theoretical foundation models and applied bioinformatics—designing AI capable of navigating unexplored biological manifolds to uncover new insights, all while maintaining rigorous safety and uncertainty awareness. 

At Novo Nordisk, Josefa spearheads the development of LLM-driven bioinformatics agents and multi-agent reasoning frameworks. Her contributions span the full lifecycle of AI deployment, from pre-training and orchestration to the engineering of evaluation harnesses. Her current research focuses on the frontiers of reinforcement learning for LLMs, specifically structure-aware reward design and uncertainty quantification to ensure model reliability in high-stakes scientific contexts. A contributor to premier venues such as ICML, ACL, and NeurIPS, Josefa is actively shaping how synthetic data and multi-agent orchestration redefine the boundaries of foundation models for science.