Feiyu Lu

Affiliated Research Scientist @Noaa National Oceanic & Atmospheric Administration

Princeton, NJ, US

Led the full lifecycle—from development to operational transition—of GFDL’s SPEAR seasonal climate prediction system, a large-scale ensemble system running on HPC infrastructure.• Designed and built a high-performance Bayesian data assimilation system for GFDL’s MOM6 ocean model, scaling algorithms for one of the world\'s fastest global DA systems.• Researched and implemented novel deep learning architectures (UNet, CNNs) to create hybrid statistical-ML data assimilation methods, published in a first-author paper.• Developed diffusion-based generative model prototypes to perform global ocean data assimilation from operational reanalysis and observational data.• Engineered and integrated Fortran-based neural network inference engines directly into production-scale global climate models, demonstrating end-to-end technology transfer from research to an operational environment.• Mentored junior researchers (undergraduate, graduate, and postdoctoral) and collaborated with internal and external research teams to publish and advance scientific discovery.

University of Wisconsin-Madison

Doctor of Philosophy - PhD

Peking University

Bachelor of Science (BS)

Peking University

Bachelor's degree