Mohammad Mahdi Kamani#
Mohammad Mahdi Kamani is a senior machine learning researcher and engineer specializing in efficient generative AI, computer vision, and distributed optimization. With years of experience developing state-of-the-art algorithms for large-scale distributed training and inference optimization, Mahdi brings deep expertise in LLMs, speculative decoding, and multimodal models. Currently serving as a Senior Member of Technical Staff at AMD’s Efficient Generative AI Team, he leads initiatives in speculative decoding optimization and multimodal model architecture. His past experiences include developing edge-cloud collaboration systems and leading multi-agent assistant systems incorporating LLM capabilities. His research contributions span federated learning, distributed training, bias mitigation, and distributionally robust models, with publications in top-tier conferences including NeurIPS, ICML, and ICLR. Mahdi holds a Ph.D. in Informatics from Pennsylvania State University, where his work focused on multiobjective optimization approaches for bias mitigation in machine learning models. He also has an M.Sc. and B.Sc. in Electrical Engineering from Sharif University of Technology.
Posts by Mohammad Mahdi Kamani
