I’m Barak Gahtan, a PhD candidate at Technion University, under the supervision of Prof. Alex Bronstein and Prof. Reuven Cohen. I build deep learning systems for real-world temporal data across wearables (physiology + activity recognition), communication networks (encrypted QUIC and teleoperation-oriented modeling), and spatiotemporal processes (differentiable eikonal solvers for wildfire propagation), with additional work on medical risk prediction from clinical records. I bring the same methods to large language models, developing diagnostics that probe how much of a model’s depth is genuine computation and whether a personalized model truly adapts to the individual. Outside the lab, I enjoy family time, coaching runners, exploring new trails, and training for my next marathon.