Summary
Riashat Islam is a Senior Research Scientist at Microsoft Research in New York with 11 years of experience advancing reinforcement learning and knowledge-rich AI systems. His PhD-trained background at Mila and collaborations with leaders like Doina Precup, Joelle Pineau, and Yoshua Bengio inform work on continuous-learning agents that maintain and predict world knowledge using probabilistic, variational, and Bayesian deep learning methods. He has a track record across industry and startups—from generative protein design and diffusion/flow-matching models at Dreamfold to alignment and multimodal agent work for SDAIA—and multiple MSR internships that shaped his applied RL expertise. Beyond core research, he blends theory and practice in preference alignment, post-training alignment, and generative modelling, and has taught graduate ML topics such as Gaussian processes and probabilistic graphical models. A less obvious strength is his sustained focus on building abstractions and representation layers across domains, enabling agents that generalize and adapt in changing environments.
11 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Machine Learning, Doctor of Philosophy (Ph.D.) Machine Learning at Mila - Quebec Artificial Intelligence Institute
Master’s Degree MPhil Machine Learning Speech and Language Technologies, Master’s Degree MPhil Machine Learning Speech and Language Technologies at University of Cambridge
University College London