Scientific Computing / Research Engineering Expert: Engineering

What you'll do

Your job is to turn engineering practice into tests for AI agents. Each task is a reproducible computational problem in a terminal: the agent gets simulation inputs, sensor data, design constraints or configuration files and has to model, optimise or debug its way to a correct result, which an automated grader then checks.

The ad spans mechanical, electrical, chemical, aerospace, civil, materials, biomedical, robotics and control systems engineering, so the breadth is deliberate. The work it lists:

• Designing multi-step tasks from real engineering and scientific workflows
• Creating datasets, simulation inputs, geometry files and sensor data
• Writing expert solutions in Python, C/C++, Julia, MATLAB/Octave or Bash
• Building containerised environments with pinned dependencies
• Writing tests that check units, physical constraints, tolerances, convergence, stability and boundary conditions
• Debugging solver, dependency and precision problems and documenting assumptions and edge cases

Who fits

• PhD, postdoctoral or equivalent advanced experience in an engineering discipline
• Strong scientific programming and comfort in Linux terminals
• Experience with simulation, numerical analysis, optimisation, signal processing or control systems
• A solid grasp of numerical methods, engineering units and technical validation

Nice to have: NumPy, SciPy, SymPy or PyTorch; tools such as OpenFOAM, CalculiX, FEniCS, ROS or LTspice-compatible workflows; FEM, CFD, robotics, embedded or digital-twin experience; Docker, Git and CI; benchmark or grader development; evaluating AI coding agents; patents or publications.

What it pays

$300 per approved task, per the ad. Review decides what counts, and the ad does not state a task quota. Compared with the sibling Earth Sciences, Mathematics and Physical Sciences listings from Turing, the stated rate is the same.

Worth knowing

Good:

• Published rate and a clear deliverable
• Wide discipline coverage, so most engineers can find a fit
• Uses real modelling and coding judgment

Less good:

• Per-task pay means unapproved work is unpaid effort
• At least 6 hours a day and 40 hours a week, with 4 hours of overlap with Pacific Time
• 5-week contract with a start expected the following week
• Contractor assignment with no medical or paid leave
• No assessment details in the ad; Turing's help centre mentions skill tests and, for some roles, live interviews

About this listing

Posted by Turing on 21 September 2026 and read on 18 October 2026. Terms are Turing's own; the comparison with sibling roles is ours. See Turing.