Computational Structural and Mechanical Engineer for AI Problem Design

Design original, graduate-level computational engineering problems that challenge advanced AI systems to perform research-level workflows using real scientific software. Problems will require running simulations, interpreting results, designing experiments, and extracting hidden information from partial data. Your work will create benchmark tasks that reveal whether AI can plan and execute complex, multi-step numerical investigations rather than just match surface patterns.

• Create challenging, original computational problems that require skilled use of domain-specific open-source scientific tools and libraries.
• Develop fully defined setups that an AI must simulate to compute exact answers, and design harder exploratory tasks where the AI must plan measurements or experiments to uncover information not directly visible.
• Write problem inputs, reference solutions, oracle functions, and automated solution validators that confirm correctness and capture nuanced failure modes.
• Run and iterate problems in remote compute sandboxes and Linux/terminal environments, testing them against state-of-the-art AI models and refining until the task reaches the intended difficulty.
• Design problems where the main challenge is reasoning and strategic planning, not sheer numerical effort, and anticipate common incorrect approaches or edge cases.
• Document problem rationale, expected workflows, and verification procedures so tasks are reproducible and testable.

• Graduate-level training in a relevant STEM field, MS or PhD preferred, or equivalent research experience.
• Proven, hands-on proficiency with at least one of the listed open-source scientific software tools, demonstrated via research publications, open-source contributions, or professional work. Relevant tools include FEniCSx/DOLFINx, scikit-fem, OpenFOAM, deal.II, MFEM, MOOSE, CalculiX, Elmer FEM, Code_Aster, SfePy, FiPy, Devito, Cantera, CoolProp, Pyomo, SimPy, or other open-source computational structural or mechanical engineering codes written in Python, C, C++, or Fortran.
• Strong Python programming skills, since you will implement problem setups, oracles, and validators.
• Solid practical understanding of numerical methods and theories such as Euler Bernoulli and Timoshenko beam theory, continuum mechanics, finite-element methods, Galerkin and variational methods, finite-volume methods, PDE discretization, constitutive modeling, numerical linear algebra, and nonlinear solution techniques.
• Experience applying these tools to workflows like beam, plate, and shell analysis, linear and nonlinear elasticity, mesh refinement and convergence studies, CFD and thermal-fluid simulation, coupled multiphysics problems, structural or system optimization, and reliability analysis is highly desirable.
• Comfort working independently, iterating on problem designs based on testing and feedback, and using Linux and terminal-based workflows with remote compute sandboxes.
• Available to work at least 15 to 20 hours per week.

Nice to Have

• Experience across multiple listed domains or tools.
• Familiarity with benchmark or evaluation design and with crafting problems or exam-style questions for scientific audiences.
• Background in scientific teaching, problem-set design, or computational reproducibility practices, including containerized environments.

• Remote, hourly engagement.
• Minimum availability requested, 15 to 20 hours per week.
• Work will be performed in Linux/terminal environments and with remote compute sandboxes provided for testing.

• Pay rate: 70 to 85 hourly.

• Must meet the qualification requirements above, including graduate-level training or equivalent research experience and demonstrable proficiency with at least one relevant scientific software library.
• Must be able to work in Linux/terminal environments and use remote compute sandboxes as part of problem development and testing.