Umar Hasan

Background

About Umar Hasan

I am an AI/ML research intern with a publication record that includes two accepted papers, with additional work under review and preparation. My work centers on building efficient, robust, and trustworthy AI for critical applications in healthcare and public safety domains. My research approach is built on three key pillars- Efficiency: Designing lightweight models for medical imaging via advanced compression techniques like Teacher-Assistant Knowledge Distillation (TA-KD), enabling deployment on resource-constrained devices- Robustness: Ensuring model reliability through rigorous cross-validation and comparative analysis, particularly for high-stakes forecasting in the humanitarian sector- Trustworthiness: Integrating Explainable AI (XAI) techniques (SHAP, LIME, Grad-CAM+) to provide transparent and interpretable insights into model predictions. My foundational interests lie in advancing NLP methodologies (LLMs, Multimodal/Multilingual Models, LCMs) and Computer Vision (Classification, Detection, Segmentation, Adaptation), with a special interest in World Models. My long-term goal is to pursue a PhD in Fall 2026 and a career in academic research, with a strong interest in the programs at KAUST, MBZUAI, KFUPM, and Khalifa University. Upon completing my current internship, I seek a Research Assistant position for Spring 2026 to deepen my experience before beginning graduate studies. I welcome connections from researchers and lab directors with ambitious projects in my areas of interest.