Earth Sciences Research Engineer
Listing checked September 23, 2026 · pay as published by Turing
Overview
Turing needs Earth Sciences specialists to build terminal-based benchmark assignments for Terminal Bench Science. You will turn real workflows from climate science, geophysics, oceanography, and related domains into self-contained tasks that AI agents must solve inside a Linux shell. Each task calls for preparing datasets, setting up a reproducible environment with scientific libraries, and writing an expert solution in Python, R, Julia, Bash, or domain software. The contract spans 5 weeks and pays per approved task.
What You'll Do10
- 1Convert genuine Earth-science workflows into standalone terminal tasks that an AI agent can attempt without outside context.
- 2Assemble datasets spanning geospatial, climate, atmospheric, geological, hydrological, or oceanographic sources.
- 3Set up repeatable computational environments stocked with the right scientific packages and command-line utilities.
- 4Write reference solutions in Python, R, Julia, Bash, or a domain-specific tool.
- 5Design exercises around geospatial analysis, time-series processing, numerical modeling, interpolation, forecasting, remote sensing, or environmental risk work.
- 6Spell out objective grading rules for scientific outputs, model behavior, data transformations, and spatial or temporal precision.
- 7Check coordinate systems, units, timestamps, missing-data handling, and the scientific assumptions behind each task.
- 8Write automated tests covering numerical tolerances, file formats, metadata, and reproducibility.
- 9Troubleshoot problems tied to geospatial projections, large datasets, dependency conflicts, performance, and numerical stability.
- 10Record input data provenance, expected outputs, edge cases, and known limitations for every task.
Requirements11
- 1Ph.D., postdoctoral experience, or comparable advanced technical background in Earth Sciences or a closely related discipline.
- 2Deep expertise in at least one area: climate science, atmospheric science, geophysics, oceanography, geology, hydrology, remote sensing, environmental modeling, or Earth-system science.
- 3Strong programming ability in Python, R, Julia, Bash, or another scientific language.
- 4Hands-on work with scientific data processing, numerical modeling, geospatial analysis, environmental datasets, or time-series analysis.
- 5Comfort operating independently in Linux and terminal-based settings.
- 6Proven ability to build, debug, and validate reproducible scientific computational workflows.
- 7Solid grasp of scientific quality control, spatial and temporal data, uncertainty, and numerical accuracy.
- 8Experience with NumPy, pandas, SciPy, xarray, rasterio, GeoPandas, Cartopy, or GDAL.
- 9Familiarity with NetCDF, HDF5, GeoTIFF, shapefiles, or GRIB data formats.
- 10Exposure to climate, weather, satellite, seismic, oceanographic, geological, or hydrological datasets.
- 11Working knowledge of Docker, Conda, Git, CI/CD, automated testing, or HPC environments.
Who Should Apply
Candidates with an Earth-science research background who write serious code and enjoy terminal work will fit this contract well. The role rewards people who already handle NetCDF or GeoTIFF files, build Conda or Docker environments, and think in terms of units, projections, and numerical tolerances. Experience in research software engineering, scientific benchmarking, or automated grader development gives you an edge, as does any history evaluating AI coding or terminal agents. You will struggle if you only have coursework-level exposure to one Earth-science subfield or if you avoid Linux-based tooling. The most common rejection reasons are tasks that cannot be verified objectively and solutions that skip edge cases such as missing data, coordinate mismatches, or unit conversions.
Salary Insight
Pay per approved task at $300 per task. Contractor assignment with no medical or paid leave. Contract duration is 5 weeks.
Pay and demand for Science & Research roles
AggregatedTypical pay
$80/hour
This role
$0.3K ≈ $0/hr
Most Science & Research roles pay $60–$98 per hour. This role's pay sits below that range.
Based on 318 similar roles that publish pay · 44 publish only a top rate; those count at the rate they gave
Rates shown per hour. Yearly and monthly pay converted; one-time fees and non-USD pay are not included.
- Live similar roles
- 381
- Listed in last 30 days
- 227
- Remote
- 92%
Hiring most right now: micro1 (151) · Mercor (53) · AfterQuery (47)
Most requested skills · share of roles
- python25%
- technical writing11%
- model evaluation6%
- docker6%
Figures from Science & Research roles live on NearSkill when this page loaded. A role can close before you apply, so check the listing itself.
Compare your resume against these rolesLocation
Compensation
$0.3K
Required Skills
Application Tip
Highlight a specific Earth-science dataset you have processed end to end, naming the exact tools and formats involved (for example xarray with NetCDF or GeoPandas with shapefiles), and describe how you validated numerical accuracy. Mention any benchmarking, grader development, or open-source contributions, because those signal the evaluation mindset this project needs.
How NearSkill helps you apply
1. Check your job match
We compare your resume with the job requirements to help you understand how well your experience fits.
2. Continue to the employer’s website
We’ll direct you to the employer’s application portal, where you can complete the application.
3. Get updates by email
If NearSkill is handling follow-up for this role, we’ll email you when there’s an update on your screening.
Want to check your fit first?
See how your resume matches this role before applying.
Check my job matchSimilar open positions
Explore active roles that match your skills and interests.
Turing
VerifiedScientific Computing Research Engineer (Remote Contract)
Turing recruits engineering experts for a frontier AI program that builds terminal-based scientific tasks used to train and evaluate AI agents. You translate authentic workflows from mechanical, electrical, chemical, aerospace, civil, materials, biomedical, robotics, and control systems engineering into reproducible computational problems. The work covers modeling, simulation, optimization, data processing, debugging, and technical validation. Expect a 5-week contractor engagement starting next week, with at least 6 hours per day, 40 hours per week, and 4 hours of daily overlap with PST.
AfterQuery
VerifiedEarth Science & Geology Expert
AfterQuery hires practicing geoscientists to create the evaluation material frontier AI labs rely on when judging models on earth and climate questions. You will design realistic scenarios that move from field observation to interpretation, then write reference answers that show the geological reasoning behind each call. The work centers on sedimentology, hydrology, geophysics, or climate science depending on your specialty, and you grade model responses for physical accuracy and calibrated uncertainty. The contract runs remote and async with a 10-hour weekly minimum, no maximum, and ongoing project work rather than one-off batches.
Turing
VerifiedPhD Physical Sciences Expert - Remote Scientific Tasks
Turing invites physical sciences experts to join Terminal Bench Science, where you convert real physics, chemistry, and materials science workflows into terminal-based benchmark tasks. You will build self-contained computational environments with pinned dependencies, write executable solutions in Python, Bash, C/C++, or Julia, and design automated tests that check numerical accuracy and physical consistency. The work covers domains such as thermodynamics, quantum mechanics, statistical mechanics, molecular dynamics, and computational chemistry. Your tasks will measure whether AI agents can reason through scientific problems, operate command-line tools, and produce trustworthy scientific artifacts. Reviewers depend on your clear notes about scientific assumptions and computational limits.
AfterQuery
VerifiedEarth, Ocean & Climate Science Researcher (Task Author)
AfterQuery builds a network of researchers who turn their own computational earth science work into evaluation tasks for AI models. You would author tasks from real research you have run and debugged, whether that is a geophysical inversion, an ocean or atmosphere model, a data assimilation step, or a remote-sensing analysis. Each task starts with a main problem, adds a few connected subproblems, and includes a Python reference solution with tests that check the result. Difficulty should come from the science, not from missing information. Pay is $200 per approved task, which takes about 2 hours, and you can take on as many tasks as you want on your own schedule. Open to research-track Bachelor's students, Master's and PhD students, postdocs, and recent graduates in geophysics, oceanography, atmospheric science, geology, hydrology, or a related field.
Turing
VerifiedMathematics Expert - Remote Scientific Computing
Turing wants mathematics experts to build authentic, terminal-based tasks for Terminal Bench Science, testing whether AI agents can handle real computational mathematics. You will craft problems across numerical analysis, optimization, statistics, probability, dynamical systems, and mathematical modeling, then implement expert solutions in Python, R, Julia, C/C++, or Bash. The work centers on translating research workflows into self-contained terminal environments with rigorous grading criteria and reproducible outputs. Every task you design must have verifiable results, clear tolerances, and automated tests that catch edge cases. This is a remote role for someone who combines deep mathematical training with hands-on scientific computing.
Turing
VerifiedAgentic Life Science Expert - Remote Computational Biology
Turing needs life sciences researchers who can turn authentic scientific workflows into structured tasks that AI agents execute from start to finish. You will build challenges where an agent inspects data, writes and runs code, picks the right computational method, troubleshoots midway failures, and lands on a defensible scientific result. Tasks live in controlled, network-isolated sandboxes, so everything from Python scripts to command-line tools and domain software must be self-contained and reproducible. Your work directly shapes how well AI systems handle real multi-step science rather than simple question answering. Expect to validate assumptions, code, and outputs, then package each task with verifiable ground truth and grading criteria.