AI for science at the Phases Research Lab — machine learning applied to physical systems, and the verification that makes the results usable.
My work is the loop that AI for science runs on: machine-learning interatomic potentials for high-throughput screening, first-principles DFT for validation, and Bayesian inference on observational data — executed on national high-performance computing systems.
hydrophonokit — a material-aware
framework for automated VASP/Phonopy phonon and thermodynamic calculations in
metal hydrides. v2.7.0 · doi:10.5281/zenodo.20337121
- Machine-learning interatomic potentials (MACE, MACE-MP), validated against DFT
- Physics-informed modelling of materials and gravitational-wave data
- Automated first-principles reference generation for training and validation
- Bayesian inference and model comparison on observational posteriors
- Verification tooling — reproducible pipelines, provenance tracking, and automated checking of published values against their source data
- Gadallah, A. A. & Liu, Z.-K. "Mechanics of Projected Horizons in Black-Hole Ringdown."
- Gadallah, A. A. & Liu, Z.-K. "Observation of Time-Dependent Kerr Parameter Trajectories in Black Hole Ringdown."
Python · MACE · VASP · Phonopy · NumPy · SciPy · SLURM · HPC
State College, PA · Open to AI-for-science, research software engineering and computational science roles. Authorized to work in the United States for any employer.