Gianni De Fabritiis
Scholar

Gianni De Fabritiis

Google Scholar ID: -_kX4kMAAAAJ
Computational Science Laboratory, Icrea Research Professor, Acellera
Computer simulationsArtificial intelligence
Citations & Impact
All-time
Citations
9,077
 
H-index
42
 
i10-index
86
 
Publications
20
 
Co-authors
159
list available
Resume (English only)
Academic Achievements
  • Pioneered research in neural network potentials (e.g., TorchMD-NET)
  • Introduced differentiable simulations via TorchMD
  • Achieved 2nd place in the Unity Obstacle Tower Challenge (NeurIPS competition) and holds SOTA performance
  • Won multiple blind challenges in drug discovery
  • Developed generative models for molecular structures (e.g., LigDream, LigaNN)
  • Created machine learning methods for 3D molecular structures (e.g., Kdeep, DeepSite)
  • Launched Donate@home—accepting Bitcoin donations to support scientific research
Research Experience
  • Co-leads the OpenMM project with John Chodera (MSKCC), Tom Markland, and Peter Eastman (Stanford University)
  • Developed ACEMD—the first biomolecular simulation software on GPUs, still one of the fastest molecular dynamics codes
  • Created GPUGRID.net—one of the world's largest distributed computing projects, pioneering GPU use in distributed computing
  • Built TorchMD—a PyTorch-based framework for molecular dynamics with seamless integration of neural network potentials
  • Developed HTMD—a Python platform for end-to-end computational biology workflows including simulation, docking, and adaptive sampling
  • Contributed to Playmolecule—an AI co-scientist application for drug discovery (now managed by Acellera)
  • Developed TorchRL—a PyTorch-based reinforcement learning library for modular agent construction