AI that survives an audit.
I take AI in the power sector from pilot to production — and make it hold up in front of a regulator.
PhD, Electrical Engineering (University of Utah) · Published in IEEE Transactions · Speaker at ISF, PMAPS, and Utility Analytics Institute
Most utility AI never makes it out of the lab.
Eighty-eight percent of AI pilots at utilities never reach production. The reason is almost never the model. It's fragmented data, workflows nobody redesigned, and a compliance question no one could answer.
Meanwhile every executive in the industry is already using these tools, and barely one in five utilities has a written policy governing them.
The bottleneck in power-sector AI is not capability. It's assurance — proving a system is reliable, documented, and defensible before it touches anything that matters.
That is the work I do.
What I do
AI readiness and value mapping
Where AI earns money in your organization, and where it quietly wastes it. A ranked map of use cases with estimated savings, sequenced by risk and feasibility.
AI governance for utilities
Use-case classification, human-in-the-loop tiers, documentation standards, and readiness for the questions a regulator will eventually ask.
AI assurance and security review
Independent evaluation of AI systems already in your stack: where they're reliable, where they fail, and how they could be manipulated.
Who I am
I build forecasting and AI systems that run in live utility operations. My PhD is in adversarial attacks on power system models — which means I spend as much time on how these systems break as on how they work.
More about me →