This page runs a Gray-Scott reaction-diffusion simulation in the background(WebGL2) described by the coupled differential equations (1) and (2). The visible webpage continuously feeds the pattern, and moving your mouse adds local disturbances. The presets are some cool behaviors I found when exploring the parameter space which you can play with yourself by adjusting the parameters below.
I'm a computational physicist with a background in software engineering. I recently completed my Ph.D. in Physics at UC Davis, where I modeled semiconductor materials and devices to understand the degradation of solar cells during fabrication.
I started my career as a software engineer, dabbling in games and educational software after earning a computer science degree from DigiPen Institute of Technology. I later returned to school to study physics and mathematics at Coastal Carolina University, before beginning my research into solar panels at UC Davis in 2020.
I enjoy investigating problems by building software and finding new ways to explore what the data reveals. I like the puzzle of trying new things, seeing what happens, and digging when the result is surprising. Making video games, and making them actually fun, taught me to build solutions that are intuitive enough to make the process quick and reliable enough to trust. Research reminded me to question everything; sometimes we should even question the question itself.
A Python implementation of coupled reaction-diffusion equations represents the Al₂O₃, poly-Si, SiOₓ, and crystalline-Si layers as connected volumes with mobile and trapped hydrogen populations. Temperature-dependent diffusion, trapping, and detrapping evolve through firing and annealing schedules using PETSc time integration. The code also runs parameter sweeps, caches results, and compares predicted trends against several experimental conditions.
This work connects atomic defect mechanisms to electrical behavior at the solar-cell scale. Quantum ESPRESSO DFT and NEB calculations examine Si–H bond breaking and hydrogen configurations. I used TCAD simulations (SCAPS-1D) to model heterojunction solar cells and understand their sensitivity to dangling-bond defects at interfaces and in doped layers. I developed Python scripts to automate sweeps over defect densities and charge states, track changes in band bending, recombination, and open-circuit voltage, and compare the results with experimental degradation trends. More details on this work can be found in my thesis, please reach out to me to request a copy of the device simulations.
A Python, LAMMPS, and Slurm workflow creates samples of amorphous SiOₓ on a layer of c-Si and generates candidate oxygen-migration paths for nudged-elastic-band calculations. It prepares the endpoint structures, runs HPC jobs, captures failed calculations, and collects energy barriers with local structural information and visualizations. The calculations showed that increased hydrogen concentration reduced oxygen diffusion, contrary to initial expectations. This indicated that oxygen diffusion alone could not explain hydrogen-induced pinhole formation.
The computational model propagates electron wavefunctions through disordered quantum-dot lattices with a transfer-matrix method, then uses finite-size scaling to study localization and the metal–insulator transition. The simulation and analysis code sweeps disorder and system size, and produces scaling fits along with other analysis.
The nreshk Python pipeline converts raw NRES telescope spectra into nightly calcium H and K activity measurements. It matches observations to calibration files, corrects and combines spectral orders, aligns spectra by cross-correlation, rejects noisy observations using autocorrelation, and extracts the activity index from wavelength windows. Diagnostic plots and consistent output let the survey combine data across stars, sites, and years.
This repository explores training a MACE interatomic potential for silicon, oxygen, and hydrogen from Quantum ESPRESSO calculations. Python scripts collect DFT inputs and outputs, extract structures, energies, and forces into MACE-compatible extended XYZ data, and provide an initial training setup. The project remains exploratory.
A C++ and CUDA implementation of the Gray–Scott model evolves two reacting and diffusing chemicals on a two-dimensional grid. The CPU and GPU backends make it possible to compare execution, while Python controls parameters and initial conditions and C++ can generate an animated GIF of the resulting spots, rings, and waves.
MieCCU extends a single-sphere scattering calculation to a lattice of metal spheres by adding the geometry and interference between scattered microwaves. The resulting simulation illustrates Bragg scattering from an ordered array.
A four-player 3D space shooter with ship combat, warping, upgrades, and sector control. I was lead designer and developed the client–server networking.
A 2D side-scrolling platform game with a custom level editor and in-game level loader. I led the design work and developed those tools.
We investigated how hydrogen diffusion causes TOPCon solar cells to lose and recover surface passivation during firing and subsequent annealing. We combined experimental measurements with a trap-limited diffusion model of hydrogen within the layers of a TOPCon solar cell to explain the non-monotonic behavior of the recombination current density, J₀, throughout this cycle. We found that hydrogen accumulates in the thin silicon oxide layer before redistributing, linking the rise and fall of J₀ to the local hydrogen concentration.
We studied how doping and illumination change the energy required to break silicon–hydrogen bonds and create recombination-active defects in both crystalline and amorphous silicon. First-principles calculations showed that p-type silicon is more susceptible to bond breaking than n-type silicon. I connected these calculations with device simulations to examine how the resulting defects affect silicon heterojunction solar cells. The simulations showed that p-type contacts also suffer greater voltage losses from these defects, helping explain their stronger degradation in experiments.
We used molecular dynamics simulations to investigate how silicon-rich pinholes form in the thin oxide layers of TOPCon and POLO solar cells during thermal processing. The simulations showed that excess hydrogen lowers the barrier for nucleation-driven phase separation, allowing pinholes to form at lower temperatures. This helps explain why hydrogen initially improves passivation but causes degradation at higher concentrations, where emerging pinholes create additional dangling-bond defects. Removing hydrogen dissolved small pinhole nuclei in the simulations, suggesting a structural mechanism for recovery.
Electron transport through a connected quantum-dot solid depends on disorder both within the dots and in the connections between them. An Anderson Hamiltonian informed by electron tomography, analyzed with the transfer-matrix method and finite-size scaling, maps the transition between insulating and conducting behavior. Weak disorder in the connections produces a crossover between distinct critical regimes over a broad part of the phase diagram.
Magnetic activity in Sun-like stars varies over years to decades and may change as stars age and spin down. The Las Cumbres Observatory telescope network makes observations publicly available after a proprietary period, providing a large archive of stellar spectra collected for many different observing programs. This survey draws on that archive to measure emission from calcium H and K lines and examine long-term activity cycles and possible changes around stellar mid-life.
Email
GitHub
Google Scholar
LinkedIn