Sebastian Junges

I am a researcher interested in the analysis of safety-critical systems.

I have been an Assistant Professor (UD) at the Software Science Group of the Radboud University in Nijmegen, the Netherlands, since September 2021. Previously, I was a PostDoc at UC Berkeley as part of the Learn and Verify Group headed by Sanjit Seshia. Before, I was a PhD student at RWTH Aachen University, under the supervision of Joost-Pieter Katoen. I defended my PhD thesis titled Parameter Synthesis in Markov Models (pdf) in February 2020.

You can reach me at sjunges?cs.ru.nl.

Research Goal

Computer science is moving fast and the automation of critical infrastructure is accelerating, also due to the advent of AI. With this power comes the responsibility to ensure that such automated processes are dependable, but our ability to certify that these systems are safe this is lagging behind. My goal is to close the gap between the automated systems we build and systems that we can certify . Challenging questions include: How can we enforce safe decision-making for drones landing in the presence of ground vehicles? Which test suites may certify that road vehicles adhere to safety standards? Answering these questions requires reasoning about system evolutions while considering the inherent uncertainty about such systems. This reasoning uses extensions of Markov decision processes (MDPs). Central to my work is developing novel verification algorithms for MDPs: These algorithms have been for network protocols, hardware architectures, railroad infrastructure, UAV decision making, etcetera. My research mostly falls within the area of formal methods and has some overlap with artificial intelligence.