What the work involves

You will be matched to one of two convection-stability areas and given an assignment agreed with you: authoring research-level problems, producing worked solutions, reviewing another contributor's work, or auditing completed items. The unit of output is not a short answer — it is a problem statement plus a derivation that a second specialist in linear stability can follow line by line and reproduce. That means stating the Boussinesq closure and the boundary conditions unambiguously, carrying the normal-mode expansion through to a solvable eigenvalue problem, and justifying any numerical truncation or spectral discretisation you rely on. Expect LaTeX for write-ups and Python/Jupyter for the numerics; familiarity with an agentic coding extension in VS Code is a plus rather than a gate, and declared gaps are accepted.

What the platform screens for

The decisive filter is publication on convection stability specifically — the listing names this as the single most common reason strong fluid dynamicists are declined. Command of Darcy's law, marginal stability analysis or spectral eigenvalue methods in some other setting is not sufficient; the screen wants your own papers on buoyancy-driven convection onset. You will be asked for three to five representative papers with DOIs or arXiv IDs, ideally from the last five years, first author strongly preferred, and every one will be checked against the public record. Journal DOIs are expected, since much of this literature never reaches arXiv. Each area you select must be backed by a specific paper of yours, with your author position and the methods it demonstrates; an unsupported selection is not scored.

Depth probes and judgment

Beyond credentials, expect follow-ups on the things that separate someone who has done this work from someone who has read about it: how the critical Rayleigh number shifts between fixed-temperature and fixed-flux thermal forcing, why the fixed-flux limit pushes the critical wavenumber toward zero, what changes when Darcy's law replaces the full momentum equation, and how you know a computed eigenvalue is converged rather than an artefact of the spectral truncation. Evaluation judgment matters too: whether a model's answer is right for the wrong reason, whether a problem you wrote has a unique defensible solution, and how you flag ambiguity rather than paper over it.

Logistics

  • Remote and fully asynchronous, no fixed hours.
  • Roughly 10 hours per week, sustained over an 8–10 week window, starting immediately.
  • Observed band of $80–110/hr, set by depth of subdomain match — stated as observed, not guaranteed.
  • Process: application form with method self-attestation and paper proofs, authorship verification, then either direct onboarding or a short live alignment call. A 30–45 minute assessment is added only where needed and most applicants will not see one.
  • Reading arXiv:2509.26574 before applying is strongly recommended; it tells you quickly whether the work interests you.