Stefano Fiorini

Research Scientist at Huawei R&D Department

About Me

I am a Research Scientist with a Ph.D. in Computer Science and a proven track record of developing state-of-the-art generative models and Geometric Deep Learning frameworks. Currently, I am conducting research on diffusion models and VAEs at Huawei R&D department, where I received the Outstanding Individual Contribution Award for technical excellence in autonomous driving innovations. My experience spans elite academic and industry R&D environments, including a current visiting position at the University of Cambridge supervised by Prof. Pietro LiΓ² and a previous postdoctoral role at the Italian Institute of Technology (IIT).

Expert in applying Deep Learning to complex domains such as high-fidelity scene synthesis, predictive modeling, and molecular interaction modeling. I have contributed foundational research to premier AI venues like ICLR, ICCV, CVPR, and AAAI, focusing on innovations such as Sheaf Neural Networks and Spectral GCNs. Beyond research, I am an active member of the scientific community, serving as a peer reviewer for major conferences like NeurIPS and CVPR and supervising younger researchers through internship and Ph.D. mentorship.

Interests: Generative Models, Graph & Hypergraph Neural Networks, Computer Vision, Autonomous Driving, and Computational Chemistry.

Diffusion Models Geometric DL GNNs Scene Synthesis Smart Mobility

Experience

2025 β€” Present Research Scientist, Huawei R&D Department Leading research on diffusion models for autonomous driving. Awarded Outstanding Individual Contribution Award.
2025 β€” Present Visiting Researcher, University of Cambridge Researching Sheaf Neural Networks and molecular modeling.
2023 β€” 2025 Postdoc, Istituto Italiano di Tecnologia Foundational research on GNNs and Diffusion Models.

Selected Publications

2026
S. Fiorini, H. Aktas, I. Duta, et al. "Sheaves Reloaded: A Directional Awakening." ICLR 2026
2025
A. Islam, S. Fiorini, S. James, et al. "Reassemble Net: Learnable Keypoints and Diffusion for 2D Fresco Reconstruction." ICCV 2025
2024
G. Scarpellini*, S. Fiorini*, F. Giuliari*, et al. "DiffAssemble: A Unified Graph-Diffusion Model for 2D and 3D Reassembly." CVPR 2024
S. Fiorini, S. Coniglio, M. Ciavotta, E. Messina. "Graph Learning in 4D: a Quaternion-valued Laplacian to Enhance Spectral GCNs." AAAI-24