I am a PhD student at the Max Planck ETH Center for Learning Systems (CLS), advised by Thomas Hofmann (ETH Zürich) and T. Konstantin Rusch (Max Planck Institute for Intelligent Systems & ELLIS Institute Tübingen).
Currently, I work on algorithmic reasoning and out-of-distribution generalization. In particular, I am interested in the theoretical foundations of learning to reason, with a focus on the question:
What makes a task hard vs. easy for an iterative reasoning model to solve?
More broadly, I am interested in dynamical-systems perspectives on deep learning and in cool connections between machine learning, topics in “classical” computer science (e.g. complexity theory) and physics.
Previously, I completed my master’s in informatics at the Technical University of Munich, where I worked with Debarghya Ghoshdastidar on the theory of self-supervised learning, and my bachelor’s in computer science at Tel-Aviv University.
Preprints
- Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
Sajad Movahedi*, Vera Milovanović*, Shlomo Libo Feigin*, Alexander Theus*, Thomas Hofmann, Valentina Boeva, T. Konstantin Rusch, Antonio Orvieto
arXiv preprint arXiv:2606.18206, 2026
*Equal contribution
- A Theoretical Characterization of Optimal Data Augmentations in Self-Supervised Learning
Shlomo Libo Feigin, Maximilian Fleissner, Debarghya Ghoshdastidar
arXiv preprint arXiv:2411.01767, 2024
Talks
- Improving Signal Propagation in Looped Transformers
Modern Numerics for Theoretical Physics seminar, Institute for Theoretical Physics, Heidelberg University · June 25, 2026
