Scientific Computing / Research Engineering Expert — Earth Sciences

This contract role is for Earth scientists who code. The work is turning real research workflows into command-line challenges that test how well AI agents handle scientific data. It suits someone with a doctorate or comparable depth in climate, geophysics, oceanography or a related discipline, and the client is an AI training company building a science benchmark.

What you'll do

• Design self-contained tasks around geospatial, climate, hydrological or other environmental datasets, with reference solutions in Python, R, Julia or Bash
• Set up reproducible environments with the scientific libraries each task needs
• Write automated checks for numerical tolerance, file formats, metadata and units
• Fix problems with map projections, dependencies, big files and numerical stability, and document data sources and edge cases

What you'll need

• A doctorate, postdoc or equivalent advanced background in the Earth sciences or something adjacent
• Solid scientific programming skills and comfort working alone in Linux terminals
• Practical experience with numerical modeling, time-series work or geospatial analysis
• Helpful extras: xarray, GDAL, GeoPandas, NetCDF, GRIB, Docker or Conda, and prior work evaluating AI agents

Good to know

• Remote contractor assignment lasting five weeks, with a start expected soon
• Requires at least 6 hours a day and 40 hours a week, including 4 hours of overlap with Pacific Time
• Pay is $300 per approved task, with no medical cover or paid leave