About
Detection you can question, not just trust.
I'm a PhD student at the Stratosphere Laboratory (AI Center, FEE CTU in Prague), working where machine learning meets network security: catching malicious behaviour in traffic, and testing whether the explanations behind those detections actually hold up.
I also design environments where autonomous agents learn to attack and defend, and build tools to reveal how those agents behave, not just the score they end up with.
Research directions
Three threads I keep pulling on.
Explanations you can measure
Detecting threats in network traffic, with rigorous metrics for the quality of the explanation behind each decision.
Autonomous attackers & defenders
Reinforcement-learning environments where offensive and defensive agents co-evolve in realistic scenarios.
How agents actually behave
Turning raw agent trajectories into interpretable characterisations of strategy: behaviour you can inspect.
Selected projects
Code and projects I work on.
NetSecGame
↗An RL environment for developing and evaluating autonomous attackers and defenders in realistic network-security scenarios.
Agent Behavior Analysis Toolkit
↗Tools for analysing the trajectories and behaviour of learned agents, turning raw rollouts into interpretable characterisations.
StratosphereLinuxIPS (Slips)
↗Open-source, ML-powered intrusion prevention system for Linux, built and maintained at the Stratosphere Lab.
Teaching & outreach
Passing it on.
Teaching Assistant, Introduction to Cybersecurity
CTU in PragueHelping students take their first steps into security, from fundamentals to hands-on practice.
Lecturer, AI Detem
Non-profitTraining sessions and workshops with AI Detem, a non-profit bringing general AI education to schools, for pupils and the teachers guiding them.