I am currently a machine learning researcher at Microsoft Research in Cambridge, UK working on foundation models for decision-making and embodied AI, in particular for data center and AI infrastructure.
Previously, I was a PhD student at the University of Tübingen and member of its Industry-on-Campus collaboration with the Bosch Center for AI. I was supervised by Prof. Gerhard Neumann from the Autonomous Learning Robots Lab at Karlsruhe Institute of Technology and received a BSc in software engineering from DHBW Mannheim and a MSc in computer science from TU Darmstadt.
Research
My journey into machine learning began when I first observed a reinforcement learning agent successfully solve a simple cartpole task. Although the task itself is not inherently challenging, it sparked my initial curiosity about the potential of machine learning agents and set me on a research path that has been both exciting and fulfilling.
My primary research focus is on advancing agentic machine learning, specifically enabling embodied AI agents to exhibit truly “intelligent” behavior. I strongly believe that this future depends on the ability to tackle complex real-world challenges and requires generalization or quick adaptation to new environments. Currently, one promising direction is to use large pre-trained models from language and vision to incorporate prior knowledge into embodied agents. While this approach is already full of interesting open questions related to training, data sources, and architectural or algorithmic design, it also provides a unique opportunity for interdisciplinary collaboration, which I find even more exciting. Although there is still a long way to go, I am excited to be a part of this ongoing journey.