Software

I develop scientific software in Python, C, Fortran, and Bash for simulation, analysis, automation, and machine learning. My work includes high-performance computing workflows using MPI, OpenMP, Slurm, GCC, NVCC, Make, HDF5, and NetCDF, alongside GPU-accelerated data generation and PyTorch-based modelling. If you are interested in working with me on a software project, please email me!

Professional / Research software

Codes developed

• committor-predictor (work in progress)
  I developed CNN-based committor predictors for nucleation in the 2D Ising model, using GPU-accelerated brute-force simulations to generate over one million labelled microstates across multiple independently trained models.
  • Constructed Markov state models along learned reaction coordinates
  • Benchmarked nucleation rates against brute-force and largest-geometric-cluster baselines
  • Automated training, validation, and MSM workflows on Slurm-based HPC clusters
  • Analysed learned models to identify physically meaningful structure in the predictions
• lammps-benchmark
  I developed an automated parameter-optimisation suite in LAMMPS for a collaborator.
  • Fully automated Python command-line interface integrated with Slurm
  • Parameter sweeps and strong-scaling benchmarks up to 128 cores
  • Log parsing and metric aggregation for reproducible comparisons
  • PDF reporting script summarising optimal parameters and scaling behaviour

Code contributions

• BraWl
  I co-developed a high-throughput Wang-Landau Monte Carlo simulation code for alloy thermodynamics and phase transitions.
  • Implemented and profiled parallel CPU performance
  • Benchmarked strong scaling to 96 cores
  • Optimised the choice of parallelisation strategy and helped finalise the 1.0 release
  • Used automated production runs on CentOS HPC clusters with Slurm scheduling for parameter sweeps
• GASP
  I contributed to the GPU-accelerated scientific computing package GASP.
  • Implemented HDF5 data storage for 2D Ising simulation datasets used in ML training
  • Developed Python CLI tools for automated CUDA-accelerated data generation, curation, and ML dataset creation
  • Built multi-GPU Slurm workflows on CentOS HPC clusters

Hobby software

Code contributions

• S.T.A.L.K.E.R. GAMMA Modpack
  I delivered Lua scripting fixes and gameplay tweaks for a video game modpack.
  • Triaged bugs and maintained backlog items
  • Partnered with a tester team to validate patches across builds
  • Shipped fixes with attention to regressions and player-facing quality