Martin Sedlacek

I'm a first year student in the ELLIS PhD programme co-advised by Josef Sivic (CIIRC) and Cordelia Schmid (INRIA Paris) working on robot learning and computer vision. I currently focus on measuring and improving the generalization capabilities of VLAs for robotic manipulation - for reference see our recent (not yet published) benchmarking work - REALM.

Previously, I completed an MSc in AI at the University of Amsterdam in 2025 with the "Cum Laude" distinction and was awarded the ELLIS Honours Research scholarship. During my time at UvA I was fortunate to work with Cees Snoek and Herke van Hoof on topics regarding Foundation Models for robotics. I also conducted research in vision-based SLAM, generative models for molecules, and reproducibility. Prior to my Masters I was a Software Engineer at Palantir, Amazon, and Goldman Sachs, spending nearly 5 years in the industry. I simultaneously obtained a BSc. in Computer Science from Queen Mary University of London (QMUL) in 2022 as part of a highly selective degree apprenticeship with Goldman Sachs. Later, I also received a GDip. in Mathematics from Birkbeck College, University of London, awarded in 2023.