Zeyu Xia

Designed and implemented a GPU-accelerated HPC platform, leveraging approximation algorithms for efficient large-scale, network-based epidemiological simulations.

Achieved 10x+ runtime improvements in complex population modeling and disease-spread forecasting, directly enhancing public health decision-making capabilities.

Architected the first GPU-accelerated, differentiable wildfire simulator, achieving millisecond-level efficiency, orders of magnitude faster than existing methods.

Developed a novel firebreak placement algorithm by integrating real-time trained diffusion and Reinforcement Learning (RL) models.

Created a decision-making tool that demonstrates robust generalization across diverse climates for effective wildfire mitigation.

Developed and deployed generative AI models (diffusion and flow matching) to precisely solve complex inverse problems in hadron spectroscopy.

Created foundational models that achieved unprecedented fidelity in correcting detector distortions, significantly advancing particle physics research.

Built reusable and FAIR-principled ML infrastructure for the MLCommons open-source ecosystem, facilitating scalable scientific discovery.

Provided comprehensive instructional support for CS 2102: Discrete Mathematics and Theory 1, enhancing student comprehension and academic performance.

Developed and deployed an end-to-end deep learning pipeline, integrating YOLOv5 and SegFormer for automated microstructural analysis.

Achieved state-of-the-art performance, outperforming existing tools and creating high-throughput capabilities for materials discovery.

Led the development of ExploreDB, a centralized data platform, effectively consolidating diverse geological datasets.

Engineered robust APIs for advanced spatial queries, enhancing data accessibility and analytical capabilities for geological research.

Designed the Magma platform to democratize big data analytics, integrating AI for scalable geospatial processing and insights.