This five-week remote contractor role involves creating reproducible terminal-based scientific computing tasks to evaluate AI agents across physics, chemistry, materials science, and astronomy workflows. Candidates need a Ph.D., postdoctoral experience, or equivalent advanced technical experience in a relevant physical-science discipline, along with strong programming skills in languages like Python, C/C++, Julia, or Bash. The position is open worldwide for English speakers and requires at least 20 hours per week with a four-hour overlap with Pacific Time.