Athar Mahmoudi

I am a Machine Learning Scientist specializing in deep learning and statistical methods for physiological time-series analysis. Currently, I lead research on non-invasive glucose estimation at Advanced Sensor Research Inc. (ASR), developing Transformer-based architectures and representation learning pipelines for noisy, real-world data.

I hold a Ph.D. in Computing Science from the University of Alberta, where I focused on Reinforcement Learning and Human-Centered AI under the guidance of Dr. Matthew Guzdial and Prof. Pierre Boulanger.

My research philosophy bridges the gap between theoretical novelty and deployment: I focus on building robust, interpretable models that generalize across domains (people and sensors) and withstand the uncertainty of applied settings. My technical core lies in sequence modeling, domain generalization, and rigorous experimental evaluation to support data-driven decision-making.