Computational Physics AI Training and Coding-Agent Evaluation Specialist DataAnnotation
Worked as an independent part-time remote contractor, completing approximately 3,600 hours across more than 1,800 specialist technical tasks in physics reasoning, computational modelling, scientific-code evaluation, model-response comparison, and reference-solution development. Designed demanding evaluation tasks spanning mechanics, fluid dynamics, electrodynamics, optics, thermodynamics, quantum physics, astrophysics, mathematical modelling, and computational methods.
Validated AI-generated reasoning and code by reviewing agent workflows, repository changes, terminal output, dependencies, numerical evidence, and debugging approaches. Assessed units, sign conventions, conservation laws, limiting behaviour, mesh and timestep sensitivity, stability, uncertainty, and agreement with independent calculations. Used Claude, Claude Code, Claude Science, Claude Cowork, Codex, and Codex CLI for scientific analysis, code review, independent cross-checking, and technical workflow evaluation.