What the work involves

You build "worlds" — self-contained accounting scenarios with enough context, source documents, and ambiguity that a strong model can plausibly fail them. A task might be a revenue recognition question under ASC 606 with a multi-element arrangement and a contract modification mid-period, a lease reassessment after a renewal option becomes reasonably certain, a purchase price allocation for a mid-size acquisition, or a SOX control deficiency that has to be evaluated for severity and aggregation. For each scenario you draft the reference output — the memo, the workpaper, the consolidated schedule, the control assessment — and then write a rubric that specifies what a competent senior would and would not conclude, and why.

The rubric is the real deliverable. Models are good at reciting the five-step model; they are much worse at deciding whether a performance obligation is distinct in context, at sizing materiality against the right benchmark, or at knowing when a technically defensible answer would still get pushed back on by a reviewer or an audit partner. Your rubrics have to make that gap legible and gradeable, with specific citations to the codification, IFRS standard, or auditing standard involved.

What the platform screens for

  • Verifiable senior ownership. Not exposure to financial reporting, but responsibility for it — engagements you signed off on, filings you owned, technical positions you defended to a reviewer, regulator, or audit committee.
  • Standards depth under follow-up. Expect the AI interviewer to take your first answer and press: which paragraph, what were the alternatives, what would the counterparty auditor argue.
  • Evaluation judgment. Whether you can articulate the difference between an answer that is wrong and an answer that is technically correct but professionally naive.
  • Track clarity. Be explicit about whether your depth is US GAAP/PCAOB, IFRS/ISA, or genuinely both. Overclaiming both tends to surface quickly.

Logistics

Fully remote and asynchronous. Contributors typically commit 10–20 hours per week, though scope varies by cohort and project phase, and some engagements run part-time over several months. Pay in the $70–80/hr range has been observed for this listing; actual rates depend on track, credential, and assessed depth, and are not guaranteed. A CPA, ACCA, CA, or equivalent is strongly preferred; prior rubric, training-material, or technical-memo authorship helps.