I am a machine learning researcher working at Cisco and specializing in neural networks and ways to exploit data structure with them. At the same time, I am doing a PhD at the Faculty of Nuclear Sciences and Physical Engineering of the Czech Technical University.
My current interests are mostly in graphs and graph neural networks, particularly graph structure and its interplay with the performance of machine learning tasks on the graph. I proposed a method for graph reduction guided by local properties of the graph. Together with my colleagues, we study the effect of graph properties on downstream tasks and their usability in hyper-parameter optimization for GNNs. Additionally, I maintain an interest in multi-instance learning, an underdeveloped field of machine learning. I am mainly interested in nested hierarchical models and their use in the field of computer network security. See more details in the publications page.
Additionally, I am an active member of the Czech Scouting organization. Since 2015, I am a lecturer at a summer course for aspiring patrol leaders, which I was also leading from 2019 until 2023. I was in previous years organising an educator conference for ca. 100 attendees as a programme chair or a general chair. I am also combining IT with scouting as a software developer for internal applications. Lastly, since 2024, I am serving as an elected member of the National Board, the top legislative and strategic body of the Czech Scouting organization.
recent publications
Convolutional Signal Propagation: A Simple Scalable Algorithm for Hypergraphs
Balancing performance and complexity with adaptive graph coarsening
