Skills

This page summarizes the simulation, coding, machine learning, HPC, and open-source tools I use most often, with examples from research, coursework, and engineering projects.

Computational mechanics

Constitutive modeling: linear elasticity, Neo-Hookean elasticity, J2 plasticity, and user material/element subroutines.
Fracture and fatigue: phase-field fracture, residual eigenstrain and thermal effects, fatigue indicator parameters, and microstructure-sensitive failure.
Numerical methods: FEM, DG, and finite-volume methods; dispersion, dissipation, and stability analysis for elastodynamics.
Structural and thermal analysis: engineering simulations for nuclear thermal propulsion and earlier turbopump work.

Simulation/CAE

LS-DYNA: used in my Graduate Research Assistant at UTK work on an ARL-funded project for high-speed impact simulations of resonant ceramic metamaterials, including Johnson-Holmquist 2 material modeling through UMAT and transient response studies. I also used ALE simulations for fluid-microstructure interaction problems. These workflows involved thousands of simulations on HPC systems, with Python, COMSOL, and LS-PrePost used for preprocessing and postprocessing.
Abaqus: used in my UTK research on an AFOSR-funded project for user-element-based phase-field modeling, including residual eigenstrain effects. I used Python scripts and Abaqus/CAE for preprocessing and postprocessing.
Ansys (Static Structural, Thermal, Material Modeller): used for structural and thermal analysis and material modeling at Standard Nuclear. Earlier experience includes Workbench, Composite PrepPost (ACP), and SpaceClaim for turbopump structural, thermal, and fatigue analyses at ITUNOVA, composite structures at Baykar and ITU Facilis, and MSc coursework.
COMSOL: used for preprocessing and mesh generation for LS-DYNA simulations in my UTK research.
SolidWorks: used for CAD preparation and engineering design workflows, especially in turbopump work at ITUNOVA and vehicle development at ITU Facilis. View my Certified SolidWorks Associate (CSWA) certification.
Siemens NX: used in broader CAD/CAE workflows for geometry cleanup, model organization, and engineering design iteration in ITU Facilis composite vehicle development.
nCode: used in MSc fracture mechanics coursework for fatigue and crack-growth analysis.
NASGRO: used in MSc fracture mechanics coursework for crack propagation and fracture mechanics analysis.

User subroutines

LS-DYNA UMAT: used in my UTK research to develop and support nonlinear material models for ceramics.
Abaqus UEL (Fortran): used in my UTK research to extend phase-field formulations to include thermal and residual eigenstrain effects.
MATLAB: used for subroutine prototyping, reduced-order model development, postprocessing, and coursework, including finite element and dynamics assignments in my MSc, finite-volume work in PhD coursework, and scripting support at ITUNOVA.

Scientific computing / in-house code

DG solvers in C++: used in my UTK research and PhD coursework to study discontinuous Galerkin formulations and compare numerical implementations. GitHub repository.
FEM solvers in C++ and Python: used for research and coursework on nonlinear finite elements, including constitutive model implementation in deal.II and related custom numerical studies.
FV solvers in C++ and Python: used for numerical method development and verification for a hyperbolic heat equation solver in PhD coursework.
Phase-field implementations: used in my UTK research for fracture and microstructure-sensitive computational mechanics studies.

Manufacturing and experiments

Component fabrication and experimental testing: manufacture and test components I design at Standard Nuclear, connecting engineering design and simulation with physical hardware.
Composite manufacturing and structural tests: workshop manufacturing and testing of vehicle structures at ITU Facilis.

ML/AI

HydraGNN: used during my Graduate Internship at ORNL for graph-based surrogate modeling workflows and code-quality contributions. GitHub repository.
PyTorch: used in my ORNL work to build graph neural network surrogates for fiber composites and ferrite-martensite polycrystals within the HydraGNN framework.
PyTorch Geometric: used in the same ORNL projects to represent microstructures as graphs and train multitask surrogate models. I am familiar with data objects, transformations, and message-passing workflows in PyTorch Geometric.
Prediction of extremes: develop graph neural networks for the maximum fatigue indicator over an entire microstructural domain, with joint grain-level and domain-level outputs. Fatigue surrogate work.
Probabilistic modeling / uncertainty quantification: used in my ORNL work for uncertainty-aware fatigue prediction and microstructure-to-property learning, including calibrated prediction intervals and extreme-value statistics linking grain-level fatigue indicators to volume-level maxima.

Research data

Open datasets: coauthor of the CP2D dataset, providing polycrystal graph representations, elastic and plastic response data, and a saved data split for reproducible surrogate-model studies.

HPC

Linux: my main environment for UTK research, ORNL internship, and MSc thesis work, including simulation, machine learning, and scientific software workflows.
Slurm: used to submit, monitor, and manage large simulation and training jobs on shared HPC clusters.
MPI: used when building and running scalable scientific software on cluster systems for larger numerical studies.
OpenMP: used in shared-memory performance settings for compiled scientific codes and related HPC workflows.
CMake: used when compiling and maintaining research software stacks, especially in the HPC and open-source workflows described in my UTK research.

Open source

MOOSE: multiphysics simulation framework.
deal.II: used in PhD coursework for nonlinear finite element implementations, including linear elasticity, Neo-Hookean response, and J2 plasticity.
OpenFOAM: used for engineering simulations at Standard Nuclear, as well as vehicle-body CFD at ITU Facilis and shock-wave studies in Gas Dynamics coursework.
LAMMPS: used in my MSc thesis for reactive molecular dynamics simulations of boron nanotubes on HPC systems.
NumPy: used throughout data preparation, postprocessing, and research scripting in simulation and machine learning workflows.
pandas: used for organizing tabular outputs and derived feature and response data in research workflows.
scikit-learn: used for baseline machine learning utilities, data preprocessing, and comparative analysis in scientific ML workflows.

Advanced Training

ATPESC 2026 - competitively selected participant, Argonne National Laboratory, July 26 - August 7, 2026.

Selected through the application process for Argonne’s intensive two-week advanced training in extreme-scale computing. Relevant program topics included:

Parallel and GPU programming: MPI, OpenMP, CUDA/HIP, and portable programming models such as SYCL and Kokkos.
Scalable numerical computing and performance: AMReX, PETSc, hypre, mixed precision, profiling/debugging, and roofline analysis.
AI and simulation workflows: distributed deep learning, LLM pretraining and post-training, inference, coupled simulation-AI workflows, and agentic tools for science.
Scientific software and data: software design, testing, verification and reproducibility, Spack, and parallel I/O with MPI-IO and HDF5.

Program agenda · Application and selection.