Mohammad Mohaiminul Islam

Generative Modeling | Geometric Deep Learning | Medical Image Analysis.

I am a PhD candidate with AMLab and qurAI, at the University of Amsterdam. Working under the supervision of Prof. Clarisa Sánchez and Prof. Erik Bekkers, we are investigating generative modeling for trajectory learning and molecular property prediction.

Before that, I worked as a researcher in EU H2020 Moore4Medical project, with the Department of Advanced Computing Sciences, Maastricht University team. I was also a Research Associate at laboratoire de traitement de l’information médicale (LaTIM) , which is a medical image processing lab under the French National Institute of Health and Medical Research. At LaTIM I particularly worked with Prof. Pierre-Henri CONZE .

As of my education, I obtained my Erasmus+ Mundus Joint Master (EMJMD) in Image Processing and Computer Vision (IPCV) from three different universities across Europe and my B.Sc. from Department of Computer Science & Engineering, Daffodil International University (DIU), Bangladesh.

News

Oct 10, 2025 Pre-print of Longitudinal Flow Matching for Trajectory Modeling is out.
Oct 10, 2025 Pre-print of Platonic Transformers: A Solid Choice For Equivariance is out.
Oct 1, 2025 Paper on Probing Equivarince and Symmetry Breaking was accepted at NeurIPS 2025.
Aug 5, 2025 Supervising M.Sc project of Coen van de Elsen, who is working with OceanOS to develop marine Foundation Model. Check out their mini-blog on medium scale training with PyTorch and Lightning.
Sep 22, 2024 Paper on Conditioning 3D Diffusion Models with 2D Images accepted to NeurIPS Workshop on GenAI4Health 2024
Jul 20, 2024 Attending summer school on Ellis Probabilistic Machine Learning at Cambridge University.
Apr 5, 2024 Paper on Segmentation via Probabilistic SDF was accepted at MIDL 2024.