I’m a theoretical particle physics PhD candidate at New Mexico State University, USA, specializing in the application of GPU-accelerated high-performance computing (HPC) and machine learning to fundamental physics problems. My work focuses on studying the intrinsic motion of quarks and gluons and exploring Beyond Standard Model (BSM) physics through large-scale simulation quantum field theory and symmetries.
My PhD research under Dr. Michael Engelhardt (NMSU) focuses on lattice quantum chromodynamics (QCD) calculations of Transverse Momentum Dependent Parton Distribution Functions (TMDs). To achieve this, I built an end-to-end machine learning pipeline to process over 30,000 multidimensional observables from Monte Carlo simulations, achieving 98%+ model fit accuracy using symbolic regression (PySR) with physics-constrained loss functions. To handle the multi-terabyte datasets generated, I developed GPU-accelerated CUDA C++ (cuFFT) pipelines, which reduced data processing time by 10x on HPC clusters. For robust analysis, I also created production-grade Python, C++, Lua and Mathematica packages to manage jackknife resampling and ensure numerical stability.
I am also collaborating with Dr. Rajan Gupta and Dr. Tanmoy Bhattacharya from Los Alamos National Laboratory on calculating the hadronic matrix elements needed to connect nucleon Electric Dipole Moments (EDMs) to BSM physics. In this role, I develop and optimize parallelized C++ CUDA kernels to accelerate multi-terabyte calculations, achieving significant runtime reductions on GPU-accelerated HPC clusters like the NERSC Perlmutter. I increased model reliability through rigorous statistical validation on over 50,000 correlated data points, applying methods like AIC-based selection and chi-squared minimization with full covariance matrices. I design and deploy custom SLURM workflows to manage and execute over 75,000 CPU/GPU compute hours, enabling robust, automated parallel analysis for these large-scale simulations.
In addition, I am collaborating with Dr. Chueng-Ryong Ji (North Carolina State University) on a project interpolating the manifestly covariant conformal group (SO(4,2)) between different forms of relativistic dynamics. For this work, I implement and manage Mathematica symbolic computation workflows on HPC clusters (NERSC Perlmutter) to analyze complex algebraic structures and symmetry constraints inherent in the problem.
My background provides me with a unique blend of deep physics intuition and hands-on expertise in C++/CUDA, parallel computing, and machine learning. I am driven to apply these skills to solve complex, data-intensive challenges and contribute to cutting-edge scientific and technical advancements.