Author materials-science coding problems — a main problem plus at least three connected sub-problems — and implement verified Python solutions with unit tests that frontier models are scored against. Work runs 40 hrs/week for 8 weeks on Turing's Central Task Platform, with 4 hours of daily overlap with PST and QC iteration until tasks clear multi-judge Pass@K criteria.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually is

SciCode tasks are not quiz questions. Each submission is a structured artifact: one main problem in materials science, decomposed into three or more sub-problems that build logically toward the final solution, plus a golden Python implementation with complete unit test coverage. The tests have to be discriminative — they must fail a plausible-but-wrong model output, not just a syntax error. Determinism matters: seeded randomness, fixed tolerances, no reliance on library version quirks or floating-point luck.

Typical subject matter sits where materials science meets computation — band structure and density-of-states calculations, elasticity tensors and Voigt notation, diffusion and Fick's law solvers, phase diagram and CALPHAD-style lever rule problems, XRD peak indexing, Monte Carlo or molecular-dynamics analysis routines, Hall–Petch and defect statistics. You will lean on NumPy, SciPy, and SymPy, and occasionally domain tools like pymatgen or ASE.

What the screen looks for

Real graduate-level materials training. Expect follow-ups that go a layer deeper than your first answer — the screen is testing whether the depth is yours or borrowed.
Test-design judgment. Can you name the specific near-miss a model would produce (transposed tensor indices, wrong unit prefix, degenerate eigenvalue mishandled) and write a case that catches it?
Discipline under rubrics. Tier 1 structure checks and Tier 2 quality rubrics are pass/fail; the target is L1 approval on first submission with low rework.
Availability honesty. 40 hrs/week and 4 hours overlapping PST for 8 weeks is a real constraint, not a preference.

Logistics

Fully remote, contractor engagement — no paid or medical leave. Eight-week contract term. Hiring is restricted to Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, and Vietnam. Work is asynchronous on the CTP, but sync calls for standups, reviews, and feedback fall inside the overlap window. Pay is undisclosed on this listing; ask for the per-task or hourly rate and the rework policy before you commit.

Scientific Coding - Material science and Python

What the work actually is

What the screen looks for

Logistics

Interview prep — 10 questions