Industrial AI
Failure detection, decision-aware model selection, industrial monitoring, and deployment-oriented evaluation.
RESEARCH
Trustworthy AI for physical and engineered systems.
My research asks when predictive performance stops being equivalent to operational usefulness — and what changes when models must operate inside real physical systems.
AI for Physical Systems
State, disturbances, controls, constraints, uncertainty, and interaction between models and physical systems.
Environmental AI
Urban water systems, wastewater treatment, climate resilience, environmental monitoring, and operational early warning.
Scientific ML
Machine learning informed by physical structure, scientific constraints, uncertainty, and mechanistic reasoning.
MODEL–REALITY GAP
A good model can still make a bad decision.
INTERACTIVE
Move across the field. The model remains fixed while the physical trajectory responds to disturbance.
STATE GAP0.00
MODELnominal
PHYSICAL SYSTEMstable
DECISIONstable