Matteo Nulli

Applied Researcher @ eBay, AI @ University of Amsterdam,

presentazione.jpeg

Science Park, 1098 XH

Amsterdam, The Netherlands

📌 Hi, this is Matteo, an Applied Researcher at eBay with the Foundation Models Team and a former ELLIS Honours student in Artificial Intelligence at the University of Amsterdam. For my master thesis, I collaborated with Yuki Asano, Ivona Najdenkoska and Mohammad Mahdi Derakhshani to investigate Compositional Reasoning in Multimodal Foundation Models. My work at eBay, focuses on training Multimodal Models and inference optimization with Vladimir Orshulevich and Hadi Hashemi.

Previously, I graduated from Bocconi University with a BSc in Mathematical and Computing Sciences for AI and was an exchange student in Applied Mathematics and Computer Science at the University of Sydney with a full-ride scholarship.

💭 My research interes lies in the intersection between Language and Vision. I am particularly interested in linking multiple modalities together. Recently I studied visual representation learning and how to leverage it for extensive image-understanding in MLLMs.

📚 Currently working at ebay on inference optimization for multimodal models as well as continuing my research into Compositional Reasoning.

Always open to hear about possible research collaborations, feel free to reach out.

news

Aug 18, 2025 Just started working as full time Applied Researcher at eBay Foundation Models Team with Hadi and Shahram. Excited to (re-)start. đŸ‘Ÿ
Apr 20, 2025 Fortunate to visit Prof. Yuki Asano at FunAI Lab in UTN Nuremberg as part of ELLIS Honours program for my master thesis on Compositional Reasoning in Vision-Language.
Mar 26, 2025 Part of Nova 111 Student List for 2025, among the most promising itaian students in my field! Read more under other stuff/.
Nov 01, 2024 Going to NeurIPS! Our paper Dynamic Vocabulary Pruning in Early-Exit LLMs was just accepted to NeurIPS Efficient Natural Language and Speech Processing (ENLSP-IV) Workshop 🚀
Jul 15, 2024 Today, I started working at eBay, joining the Foundation Models Team in the Amsterdam office. The internship will focus on multimodal research and development for data, architecture and pre-training. Excited to start!
Jul 03, 2024 In-Context Learning Improves Compositional Understanding of Vision-Language Models was just published in ICML Workshop on Foundation Models in the Wild 🙌
Jun 01, 2024 My first paper ‘Explaining RL Decisions with Trajectories’: A Reproducibility Study was just published in Transactions on Machine Learning Research 🎊
Sep 01, 2023 I joined the University of Amsterdam to pursue the Master in Aritifcial Intelligence
Jul 25, 2023 I graduated from Bocconi University and was among the first 9 graduates in the history of the course Mathematical and Computing Sciences for AI! 🎓
Feb 10, 2023 Just landed in Australia to join the University of Sydney for a full semester, I will be taking Deep Learning, Stochastic Processes and Big Data and Data Diversity courses as well as surfing 🏄

selected publications

  1. Dynamic Vocabulary Pruning in Early-Exit LLMs
    Jort Vincenti*, Karim Abdel Sadek*, Joan Velja*, Matteo Nulli*, and Metod Jazbec
    NeurIPS Efficient Natural Language and Speech Processing, 2024
  2. In-Context Learning Improves Compositional Understanding of Vision-Language Models
    Matteo Nulli, Anesa Ibrahimi, Avik Pal, Hoshe Lee, and Ivona Najdenkoska
    In ICML 2024 Workshop on Foundation Models in the Wild , 2024
  3. ’Explaining RL Decisions with Trajectories’: A Reproducibility Study
    Karim Abdel Sadek*, Matteo Nulli*, Joan Velja*, and Jort Vincenti*
    Transactions on Machine Learning Research, 2024
    Reproducibility Certification