projects

Research projects I'm currently working on.

research

  1. Degeneration Detection Is Not Early Warning
    Marco De Negri, Luca Sartori, Lorenzo Baggi, Moritz Reihs, Eduard F. Ďurech, Mennatallah El-Assady, Valentina Pyatkin, and Anna Hedström
    ETH AI Center, Apertus Group, Spring 2026
    Linear probes on LLM hidden states that warn, token by token, before a model falls into a repetitive loop. Accepted at the NeurIPS 2026 Workshop on Interpretability as a Science.
  2. Generative Antibody Design with Protein Language Models
    Marco De Negri, Giovanni Guidarini, Andreas Krause, and Scott Sussex
    Learning & Adaptive Systems Group, ETH Zürich, Spring 2026 – present
    Fine-tuning ESM2 with Direct Preference Optimization to generate higher-affinity antibody variants.
  3. Goal-driven 3D Map Generation with Reinforcement Learning
    Lorenzo Baggi, Marco De Negri, Vassili De Palma, Alessio Salvatore, Daniel Barath, and Jelena Trisovic
    Computer Vision and Geometry Lab, ETH Zürich, Spring 2026
    An RL agent that dynamically allocates resolution in a 3D map under a tight memory budget.