KDD · 2026
Cosmo3DFlow
Flow matching with 3D wavelet representations for cosmological reconstruction: up to 46x faster sampling than diffusion models on 128^3 simulations (KDD 2026).
I am a Computer Science PhD student at the University of Virginia, advised by Prof. Geoffrey C. Fox. My research focuses on generative models and differentiable simulation for science. I build efficient methods for scientific data generation and inverse problems, and evaluate how well models capture physical behavior.
I also contribute to DeepVELA, exploring agentic benchmarks for scientific applications in collaboration with MLCommons.
- Generative Models
- Differentiable Simulation
- Scientific Model Evaluation
Selected Work
IEEE IRI · 2026
FireDataForge
FireDataForge is an open-source Python framework that automates retrieval and harmonization of 11 wildfire-related geospatial data sources, producing analysis-ready arrays for fire behavior simulation, machine learning, and AI-assisted wildfire analysis.
Journal of Instrumentation · 2026
ScatterPrism
Conditional flow matching for particle simulation and inverse problems, with multi-metric evaluation of physical fidelity beyond training loss.
Environmental Modelling & Software · 2025
PyTorchFire
GPU-accelerated differentiable wildfire simulation in PyTorch, with gradient-based parameter calibration to observed fire spread.
Get in touch
I am seeking 2027 research internships in generative AI and ML systems.
Email me to discuss research opportunities or collaborations.