Marta Grobelna

Doctoral Researcher in Algorithm Development at the Technical University of Munich

prof_pic.jpg

I am a doctoral researcher at the Technical University of Munich, supervised by Prof. Jan Křetínský. I design algorithms for decision-making under uncertainty, with formal guarantees on how precise their answers are, and I turn them into working software.

My research focuses on concurrent stochastic games and Markov decision processes, which model systems where several agents act at the same time and chance plays a role. I developed the first value-iteration algorithm for concurrent stochastic games with guaranteed precision (LICS 2025) and implemented it in the PRISM model checker, using AMPL and Gurobi to solve the underlying optimization problems. You can read more on the project page.

Alongside my research, I lead the modernization of Automata Tutor, a web platform that grades exercises in automata theory automatically and has been used by thousands of students worldwide. I rebuilt its frontend in Python/Flask, moved it to a new database and set up its testing and monitoring.

Before my PhD, I worked at the Fraunhofer Institute for Cognitive Systems (IKS) on industry projects with BMW and Huawei. There, I turned questions about the safety of autonomous driving into precise mathematical problems and delivered software prototypes.

I will finish my PhD in early 2027. Next, I want to bring this experience to industry and tackle challenging real-world problems. Feel free to get in touch or have a look at my CV.

Outside of work, you will find me bouldering, cycling, hiking in the mountains or playing guitar.

selected publications

  1. LICS
    Stopping Criteria for Value Iteration on Concurrent Stochastic Reachability and Safety Games
    Marta Grobelna, Jan Křetínský, and Maximilian Weininger
    In 40th Annual ACM/IEEE Symposium on Logic in Computer Science (LICS), 2025
  2. EDCC
    Dynamic Risk Management for Safely Automating Connected Driving Maneuvers
    Marta Grobelna, Joao-Vitor Zacchi, Philipp Schleiss, and Simon Burton
    In 17th European Dependable Computing Conference (EDCC), 2021