Studies of electrical engineering and information technology at the Karlsruhe Institute of Technology (KIT). Bachelor thesis at the Forschungszentrum Informatik (FZI), developing a driver assistant function for optimizing the energy efficiency of the traffic flow (2016). Internships at TRUMPF Inc. in Farmington, CT (USA) on optimizing the control of laser cutting machines and at BOSCH SEA Pte. Ltd. in Singapore contributing to the core development of an Ubuntu Linux based operating system. Master thesis at the Institute of Control Systems (IRS), developing an intention based cooperative control concept for human-robot-interaction with a variable level of automation (2019).

Doctoral Researcher at IRS since March 2020, leading the Functional Safety Control research group since April 2024. 

Autonomous systems promise substantial benefits to society. In almost every relevant application, however, they are safety-critical, and therefore the potential benefits only materialize if safe operation can be guaranteed rather than merely observed in testing. This holds equally for autonomous systems that employ AI-based components and for those built from classical components only.

My research addresses this gap by developing methods for provably safe motion generation. The central concern is execution accuracy: given the potential unpredictability of AI-based controllers, external disturbances, and inaccurate modeling, how closely can a system be guaranteed to follow its intended motion? This leads to the question that drives my work — how conservatively must a system behave, for instance how slowly it must move, in order to guarantee a required accuracy? Answering it turns safety from a binary property into a quantifiable trade-off between performance and guarantee.

I develop the associated methods on a game-theoretic basis, within a framework that deliberately leaves room for learning-based components: AI-based controllers may act freely within a verified operating envelope, while a formally verified safeguard intervenes only when a guaranteed accuracy — and with it, for instance, freedom from collisions with obstacles in the system's surroundings — would otherwise be lost