Summary
Amit Hasan is a PhD candidate and Graduate Research Assistant at the University of Connecticut specializing in efficient machine learning systems, with 11 years of experience applying algorithm–system co-design to make large models and graph neural networks practical on constrained hardware. His work spans quantization for deploying LLMs on small GPUs, GNN pruning and SIMD-aware kernels, and a GPU accelerator for GCNs, producing multi-fold runtime and memory improvements that have been recognized at DAC, ICCAD, ASPLOS, and HPCA. Prior to his PhD he shipped production ML solutions at Hiperdyne in Tokyo, improving user-group classification and anomaly detection for Sony’s MVNO, and built reinforcement-learning agents for research competitions. He combines deep theoretical insight with hands-on engineering—evident in a submitted ICLR paper and conference-grade system designs—while also applying vision models to real-world problems like traffic safety and disaster damage assessment. Outside research he pursues robotics, electronics, photography, and continuous self-directed learning, often prototyping ideas end-to-end from model to deployment.
11 years of coding experience
3 years of employment as a software developer
SSC, Science, SSC, Science at Polli Unnayan Academy Laboratory School and College
Bachelor of Science, Electrical and Electronics Engineering, Bachelor of Science, Electrical and Electronics Engineering at Bangladesh University of Engineering and Technology
HSC, Science, HSC, Science at Bogra Cantonment Public School and College
Bengali, English, Japanese