Arnab Kumar-Mondal

Arnab Kumar-Mondal is a Machine Learning Researcher at Apple Inc., with a Ph.D. in Deep Learning from McGill University and Mila – Quebec Artificial Intelligence Institute. His work bridges theoretical and applied research in computer vision, language modeling, robotics, and AI for science.

• Ph.D. from McGill University (2025 completion)
• Internships at Microsoft Research and Apple
• Visiting Researcher at ServiceNow Research and Huawei Noah’s Ark Lab
• B.Tech in Electronics and Electrical Engineering from IIT Kharagpur

His research focuses on equivariant learning, state space modeling, and generative adversarial networks (GANs), with applications in medical imaging, human motion analysis, and vector graphics generation. Key contributions include canonicalization frameworks for symmetry-aware modeling and spectral analysis of representation quality in self-supervised learning. Collaborations span institutions like ServiceNow, Huawei, and Mila.

Recent publications (2023–2025) explore symmetry-aware generative modeling, efficient dynamics modeling in interactive environments, and rotation-invariant visual representation learning. His work on ternary language models at ICLR 2025 demonstrates scalable pretraining techniques.

Arnab maintains active contributions to open-source software, including PyTorch implementations for semi-supervised segmentation via CycleGAN. His technical depth extends to VLSI engineering, embedded systems, and free-form lens design from undergraduate research.

Professional activities include patents on video-language foundation models, internships at leading tech firms, and cross-institutional research roles.