Marco Fumero

About

Marco Fumero is an incoming Kempner Research Fellow at Harvard University and a Marie Skłodowska-Curie Postdoctoral Researcher at the Institute of Science and Technology Austria (ISTA), where he works with Francesco Locatello. He received his Ph.D. in Computer Science from Sapienza University of Rome under the supervision of Emanuele Rodolà and was part of the ELLIS PhD program, during which he completed research internships at the Autodesk AI Lab and at Amazon AI Research (AWS). His research has focused on representation alignment, causal representation learning, and representational geometry, with an emphasis on understanding shared structure across learned representations. In line with these interests, he is an organizer of the UniReps workshop series, on unifying representations in neural models.

Research Focus

Marco’s research lies at the intersection of representation learning, geometry, and causal learning. It seeks to understand when and why machine learning models trained with different architectures, objectives, and modalities learn similar representations, and what this convergence reveals about the mechanisms underlying learning and generalization. By analyzing the geometry of learned representations and their relationship to causal factors and data invariances, Marco develops methods for comparing and aligning representations across models, enabling efficient transfer and reuse of information across pretrained systems. More broadly, this perspective aims to make learning systems more robust, reliable, and interpretable, while uncovering universal organizing principles of representations that can ultimately inform mechanistic understanding of learning in both artificial and biological systems.