Umar Hasan

About

I build explainable, efficient AI systems for healthcare and safety-critical applications.

My research focuses on making deep learning models both interpretable and deployable in resource-constrained settings. I've developed lightweight diagnostic frameworks using knowledge distillation and PEFT, achieving competitive performance with models up to 94% smaller than traditional baselines, applied to medical imaging, physiological time-series, and safety prediction.

Currently collaborating with researchers at Duke University on multimodal foundation models and at North South University on vision-language models. My work has been published or is under review at venues such as Sensors (IF 4.0), Diagnostics (IF 3.8), IEEE Access (IF 3.6), NeurIPS, and ICML.

I'm seeking Fall 2026 or Spring 2027 MS/PhD opportunities to contribute to ambitious research at the intersection of trustworthy AI and real-world impact.

Research Lab Experience

Publications

My research focuses on medical foundation models, AI-assisted clinical decision support, and model interpretability. To avoid the overhead of manual tracking, a continuously updated list of my peer-reviewed articles, preprints, and citation metrics can be found directly on my Google Scholar Profile.

Honors & Awards

Teaching Experience

Professional Service

Selected Software Project

Technical Skills

Languages