
- Researcher | Engineer
Hello! My name is Ehimare and I am a machine learning researcher completing my PhD at the Professorship of Energy Management Technologies at the Technical University of Munich. My research asks when a learned model of a physical system can be trusted, particularly outside the conditions it was trained on. In the past I have studied this question primarly in the power systems domain but have now expanded this to include the Earth system at large.
My research currently revolves around:
(a) Machine learning for weather and climate prediction, power systems modeling, and decentralized energy systems
(b) Generalization under distribution shift, including Graph Neural Networks and their theoretical properties
(c) Physics-informed and hybrid models that embed physical knowledge into learned components
(d) Uncertainty quantification and reliable machine learning
Climate and energy each raise hard, important questions in their own right, and both must be understood, predicted, and steered reliably as we move away from fossil fuels. My approach in each case is to use machine learning to accelerate computation, fill gaps in our physical models, and uncover new insights from data, while keeping the results reliable enough to be trusted. I like to work where theory meets practice, and my ultimate goal is to help tackle climate change and accelerate the clean energy transition.