The work
You will turn real Earth sciences and computational research workflows into terminal-based tasks for training and evaluating AI systems. Each task should require meaningful scientific reasoning, code execution, troubleshooting, and objectively verifiable results.
- Design multi-step tasks covering climate science, atmospheric processes, geophysics, oceanography, geology, hydrology, remote sensing, and environmental modeling.
- Prepare and process climate, weather, atmospheric, oceanographic, geological, hydrological, satellite, and other environmental datasets.
- Develop geospatial analysis, time-series processing, numerical modeling, interpolation, forecasting, remote sensing, and environmental risk tasks.
- Build reproducible terminal environments with scientific libraries, command-line tools, and required dependencies.
- Create reference solutions with Python, R, Bash, Julia, or relevant scientific software.
- Write automated tests and evaluation criteria for scientific correctness, numerical accuracy, spatial and temporal consistency, and reproducibility.
- Check coordinate reference systems, units, timestamps, missing data, uncertainty, and scientific assumptions.
- Document data sources, workflows, expected outputs, edge cases, and scientific limitations.
What it pays and takes
This is a remote contractor assignment open worldwide. The work requires advanced Earth sciences knowledge, scientific programming, and experience with Linux and terminal-based computational environments.
- Pay information is not provided in the listing.
- 20+ hours per week.
- Minimum four-hour daily commitment and four hours of overlap with Pacific Time.
- Fluent English.
- Ph.D., postdoctoral experience, or equivalent advanced technical experience in Earth Sciences, Climate Science, Atmospheric Science, Geophysics, Oceanography, Geology, Hydrology, Environmental Science, or a closely related field.
- Strong programming ability in Python, R, Bash, Julia, or another relevant scientific language.
- Practical expertise in scientific data processing, numerical modeling, geospatial analysis, climate modeling, remote sensing, or environmental data analysis.
- Strong understanding of spatial datasets, time-series analysis, numerical accuracy, uncertainty, data quality control, and reproducible workflows.
- Ability to independently develop, debug, validate, and clearly document computational solutions.
- Helpful tools and formats include NumPy, pandas, SciPy, xarray, rasterio, GeoPandas, Cartopy, GDAL, NetCDF, HDF5, GeoTIFF, shapefiles, GRIB, Docker, Conda, Git, CI/CD, and automated testing frameworks.
- Additional useful experience includes high-performance computing, Earth-system models, environmental forecasting, geophysical simulations, research software engineering, scientific benchmarks, automated grading, AI coding agents, or scientific quality control.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind systems that learn from examples. In this role, your scientific tasks, reference solutions, and evaluations help measure whether AI can produce reliable computational results in Earth sciences.
People with advanced technical knowledge are needed because they can judge scientific assumptions, numerical accuracy, reproducibility, and the limits of model-generated work.