All Pioneers (100)
Profile 17 of 100
1993– advanced

Priya Donti

Assistant Professor of EECS at MIT & Co-Founder of Climate Change AI

Priya Donti

Biographical Overview

Embeds hard physical laws and electrical grid constraints directly into neural networks so they cannot produce dangerous commands. Donti co-founded Climate Change AI and develops optimization-constrained machine learning that helps electric power grids safely absorb high levels of renewable energy.

"Machine learning is not a silver bullet for climate change, but when fused with physical principles and domain constraints, it provides crucial tools to decarbonize the power grid and accelerate the clean energy transition."

— Priya Donti
Lifespan 1993–
Technical Depth advanced
Key Breakthrough Constraint-Enforced Neural Networks (DC3) for Electric Grid Decarbonization
Focus Areas
physics-informed ML power systems constrained optimization climate technology
Topic Keywords
#climate-AI #power-systems #constrained-optimization #physics-informed-ML #deep-learning #implicit-layers #renewable-energy
Source: Historical Biographical Archive / Wikimedia Commons
💡

Historical Context & Impact

In short

Standard neural networks are black boxes that frequently output physically impossible recommendations. Donti's algorithms mathematically embed power-flow physics directly into the forward pass of the model, guaranteeing that the AI cannot issue an instruction that overloads a transmission line or triggers a blackout.

Key Technical Breakthroughs & Inventions

01
Optimization-Constrained Deep Learning (DC3) Developed Deep Constraint Completion and Correction (DC3), an architecture that enforces physical grid limits directly within neural network outputs, solving non-convex power problems up to 100x faster than traditional numerical solvers without breaking physical bounds.
02
Tackling Climate Change with Machine Learning Co-authored the landmark taxonomy mapping how machine learning can accelerate power grid decarbonization, industrial efficiency, and climate modeling, establishing a new global research discipline.
03
Safe Grid Inverter & Frequency Control Formulated predictive reinforcement learning controllers for renewable energy inverters, helping solar- and wind-heavy microgrids stay stable without the physical rotational inertia of fossil fuel turbines.
04
Co-Founding Climate Change AI Co-founded Climate Change AI (CCAI) to fund research, organize top-tier academic workshops at NeurIPS and ICML, and connect machine learning researchers directly with electric grid operators.

Selected Honors & Industry Recognition

Original Publications, Papers & Archives

Connected Contemporaries

All 100 Pioneers →