AI / ML for climate resilience, wind forecasting, and SDG-aligned infrastructure

I design physics-informed forecasting systems for clean energy, grid reliability, and city-scale thermal-stress early warning. My work targets operational resilience under real hardware limits, including NVIDIA Jetson Nano–class devices and other edge form factors with strict memory and latency budgets.

In many sites, the core objective is to forecast dangerous behavior before it cascades: extreme wind spikes that threaten grid stability, or localized heat stress that endangers vulnerable neighborhoods. I build models that prioritize those extremes instead of just average accuracy. These models often reduce high-risk error by more than 50% compared to standard deep-learning baselines while still fitting on constrained hardware.

This work aligns with Sustainable Development Goal (SDG) 7 Affordable and Clean Energy through high-fidelity short-horizon wind speed prediction, SDG 11 Sustainable Cities and Communities through neighborhood-scale urban heat-risk early warning, and SDG 13 Climate Action through adaptation-oriented extreme-event forecasting.