Controlled BSk24 sound-speed deformations
Methods and software · 2026–present
Developed a fail-closed workflow for smooth sound-speed modifications, effective cold-barotrope reconstruction, and TOV/tidal response calculations. Released versioned software and checksum-verified campaign evidence; sole-authored public arXiv preprint arXiv:2608.23033v1.
Machine Learning Classification of Neutron Star Composition
Undergraduate thesis · examined February 2026
Built a synthetic equation-of-state, stellar-structure, and machine-learning workflow to study restricted compact-star model discrimination. A subsequent audit narrowed the defensible classification scope and documented leakage, provenance, and generalization risks.
Post-thesis classifier audit and interactive demonstrations
Methods and software · 2026–present
Audited and redeveloped the restricted synthetic classification workflow, with physical-family partitioning and explicit interpretation limits. Developed separate baseline and perturbation-sensitivity demonstrations whose outputs are labelled as uncalibrated model scores rather than astrophysical probabilities.