Kazi Ashik Islam

Kazi Ashik Islam has joined the Learning Systems Group at ORNL as a Postdoctoral Research Associate in 2026. His research is currently focused on designing and formalizing sub-quadratic attention mechanisms with structured hierarchical memory, targeting efficient training and sublinear-memory decoding.

Kazi Ashik Islam holds a PhD in Computer Science from the University of Virginia, where his research focused on computational tools for public policy and decision-making, particularly in large-scale disaster management and infrastructure development. His work combines machine learning, optimization, and agent-based modeling to solve complex real-world problems. His research works have been published in venues including AAAI, PNAS, IJCAI, CIKM, and IEEE conferences.