Jae-Won Chung

I am a final-year PhD candidate in CSE at the University of Michigan, advised by Professor Mosharaf Chowdhury. I build efficient software systems for machine learning, with a focus on the efficient management of not only time, but also energy and power. I view energy as a first-class systems resource that is worth carefully optimizing and allocating across the full stack based on precise understanding and control.

Building real systems and tools to produce impact across open-source communities and industry is a key part of how I do research, with many of these systems and tools developed in the context of The ML.ENERGY Initiative that I founded and lead. Specifically, the open-source Zeus project is part of the PyTorch Ecosystem, where it has active real-world users and contributors; Perseus, Kareus, and the ML.ENERGY benchmark have been adopted by industry including NVIDIA and the MLPerf Power benchmark; and my work has influenced broader discussions on AI energy consumption through major outlets including the MIT Technology Review and the G7 French Presidency.