Trustworthy AI for thermal systems.

We build evaluation and prediction tools that show when thermal AI models for data center liquid cooling can be trusted, and when they cannot.

What we do

Physics-aware similarity checks that detect train/test leakage in boiling and heat transfer datasets.

Benchmarks that hold out whole facilities, fluids and surfaces to test whether models hold up outside their training data.

Critical heat flux prediction with calibrated uncertainty and out-of-distribution warnings for two-phase and liquid cooling.

An open benchmark for evaluating machine learning models of boiling heat transfer.

Calibrated critical heat flux prediction with out-of-distribution warnings.

Open source code supporting this work.

Founder and Principal Investigator

BS, Mechanical and Energy Engineering, University of North Texas. Certified Energy Manager.

Operations and Research Associate

BCP Technologies LLC, Hutto, Texas