Hossein Abdi

I am a final-year Ph.D. researcher at The University of Manchester, in the Department of Computer Science (Machine Learning and Robotics group). My research focuses on developing scalable high-order policy optimization algorithms in reinforcement learning with a novel Kalman-based Bayesian approach. My work provides theoretical convergence analysis, with large-scale policy training on GPU clusters using PyTorch/JAX, and practical deployment on Franka Panda robotic arm. I am particularly interested in decision-making and control using large-scale Foundation Models.