The work

Two halves. First, authoring: you take work you've actually done — an operating budget build, a month-end flux analysis with commentary, a daily cash position, a debt covenant compliance schedule — and reconstruct it as a scenario with enough context that a model can attempt it and be graded fairly. That means specifying the inputs, the judgment calls, and what a defensible answer looks like, including the assumptions a competent analyst would flag rather than silently accept.

Second, evaluation: you review model outputs side by side and rank them. Most of the signal here isn't arithmetic — it's whether the model chose a sensible forecast driver, handled a timing difference correctly, understood the difference between cash and accrual in a 13-week forecast, or wrote variance commentary that explains cause rather than restating the number. Feedback is written prose, not a score, and researchers read it.

What the screen looks for

  • Concrete artifacts you've owned end to end, with company scale and reporting cadence attached
  • Depth under follow-up: expect probing on FX revaluation mechanics, covenant definitions, bank administration, or model structure until you either demonstrate or run out of depth
  • Ability to say why one output is better, in writing, without hedging
  • Real availability — vague commitments read as noise

CTP helps for treasury-leaning applicants; it isn't required. Corporate FP&A at a mid-size or larger company matters more than brand name.

Logistics

Fully remote and asynchronous, contract, no fixed hours. Contributors typically commit 10–20 hours weekly, though scope varies with project phase and some engagements taper without notice. Application is resume plus a short practice-area form; shortlisted candidates get a sample task before any pay begins.