I am a Senior Researcher at NEC Labs Europe. My research interests include Machine Learning, Optimization, Geometry. I am working at the intersection of Physics and Machine Learning, in particular modelling biomolecular systems. I am part of the Machine Learning for Computational Science team at the Interlligent Software System Group.
Our paper on learning unbounded layer width during training (with easy truncation and dynamic compression) has been accepted for poster presentation at ICLR 2026, Rio de Janeiro.
Our paper on the Long-Range Ising Model benchmark for provably evaluating long-range capabilities in graph learning has been accepted for poster presentation at ICLR 2026, Rio de Janeiro.
Talk at the Workshop on Principles of Generative Modeling (PriGM), EurIPS 2025, Copenhagen. We introduce a formal logical composition framework for conditional diffusion (AND and AND-NOT guidance) and demonstrate it for antibody CDR design.
At the 1st SIMBIOCHEM Workshop (Dec 6th, 2025, Copenhagen), Viktor Zaverkin presented our work on augmenting MLIPs with analytic long-range electrostatics and dispersion for transferable molecular and bulk simulations.
Performance of Universal Machine-Learned Potentials with Explicit Long-Range Interactions in Biomolecular Simulations accepted for publication in JCTC (ACS). Viktor Zaverkin, Matheus Ferraz, Francesco Alesiani, Mathias Niepert.
All news and updates.
Our paper on learning unbounded layer width during training (with easy truncation and dynamic compression) has been accepted for poster presentation at ICLR 2026, Rio de Janeiro.
Our paper on the Long-Range Ising Model benchmark for provably evaluating long-range capabilities in graph learning has been accepted for poster presentation at ICLR 2026, Rio de …
Talk at the Workshop on Principles of Generative Modeling (PriGM), EurIPS 2025, Copenhagen. We introduce a formal logical composition framework for conditional diffusion (AND and …
At the 1st SIMBIOCHEM Workshop (Dec 6th, 2025, Copenhagen), Viktor Zaverkin presented our work on augmenting MLIPs with analytic long-range electrostatics and dispersion for …
Performance of Universal Machine-Learned Potentials with Explicit Long-Range Interactions in Biomolecular Simulations accepted for publication in JCTC (ACS). Viktor Zaverkin, …
Fast, Modular, and Differentiable Framework for Machine Learning-Enhanced Molecular Simulations accepted in J. Chem. Phys. PyTorch-based framework for ML-enhanced MD/MC; code at …
Added 2025–2026 papers from Google Scholar and upgraded the site to Hugo 0.155 with the latest theme.
