Dr.
David Berghaus

David Berghaus studied physics at the University of Bonn, where he received a BCGS Honors Branch Scholarship. From 2021 to 2023, he completed his PhD in Theoretical and Mathematical Physics at the Rheinische Friedrich-Wilhelms-Universität Bonn, graduating magna cum laude. His dissertation, On the Numerical Computation of Modular Forms, contributed to advances in computational number theory.

Since 2023, he has been working as a Postdoctoral Research Scientist at the Lamarr Institute for Machine Learning and Artificial Intelligence and simultaneously as an Applied Research Scientist at the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS). His work spans both fundamental research and applied industrial projects in the fields of time series analysis and natural language processing.

At the Lamarr Institute, he investigates the application of deep learning for modelling stochastic processes. His recent work includes a foundation inference model for Markov jump processes, accepted at NeurIPS 2024, as well as research on stochastic differential equations and point processes. His broader interests lie in combining statistical learning and mathematical modeling to better represent and manage uncertainty in real-world systems – particularly in the natural sciences.

In parallel, at Fraunhofer IAIS, he develops large language model (LLM)-based solutions for industrial use cases. His projects include fine-tuning LLMs for legal tasks in German lawand designing production-ready systems for the automated processing of structured email orders. His work integrates modern NLP architectures, task-specific adaptation, and workflow-oriented deployment strategies.