Published multiple papers including preprints like 'How does the optimizer implicitly bias the model merging loss landscape?' and conference/journal publications such as 'In Search of Adam’s Secret Sauce', NeurIPS 2025 (Oral). Awarded Schmidt Sciences AI2050 Early Career Fellow.
Research Experience
Principal Investigator (PI) at ELLIS Institute Tübingen and independent group leader at the MPI for Intelligent Systems. Leads the Deep Models and Optimization group, lectures at the University of Tübingen, and is faculty for CLS, ELLIS, IMPRS-IS PhD Programs. Interned at DeepMind London, Meta (FAIR) Seattle, MILA, and Inria Paris. Involved in computational systems biology projects at ETH, such as SignalX, and regularly helped in rare diseases research with bioinformatic analysis of genome sequence data from EEC syndrome patients.
Education
PhD from the Data Analytics Lab at ETH Zürich, supervised by Prof. Dr. Thomas Hofmann and Dr. Aurelien Lucchi. Master's degree in Robotics, Systems, and Control from ETH Zürich.
Background
Research interests include improving the efficiency and capabilities of deep learning technologies in science and engineering by pioneering new architectures and training techniques. Main research areas are understanding large-scale optimization dynamics and designing innovative architectures capable of reasoning with complex data. Focus is on exploring innovative techniques for decoding patterns in complex sequential data.
Miscellany
Enjoys traveling and reading philosophy books in his free time. Favorite authors include Kierkegaard, Lévi-Strauss, Meister Eckhart, and Nietzsche. Plays several instruments, including cello (studied for over 10 years in Venice and Klagenfurt), currently learning oboe, and occasionally plays transverse flute and acoustic bass.