What the work actually involves

You will spend most of your time doing two things. First, authoring reference artifacts: a driver-based operating model with clean assumption tabs, a variance bridge deck that walks budget to actual by driver, a monthly flux commentary memo that a CFO could forward untouched. Second, grading what the model produces against that standard — marking where a forecast formula breaks when a driver changes, where a bridge waterfall doesn't tie, where a commentary memo describes a variance without naming its cause. Written rationale matters as much as the verdict; the lab is training on your reasoning, not just your score.

Tasks arrive across the workflows named in the listing: rolling forecast models, variance bridge decks, driver-based operating models, board-ready P&L decks, monthly flux commentary memos, and scenario sensitivity models. Expect a mix of file formats — spreadsheets with live formulas, slide decks with actual layout, and prose memos. The craftsmanship bar is the same one your finance leadership applied: does it tie, does it hold up under a follow-up question, would you present it.

What the platform screens for

  • Depth over titles. The screen probes specific builds you owned — how you structured drivers, what you did when actuals broke the model, how you handled a bridge that wouldn't reconcile.
  • Evaluation judgment. Can you distinguish a genuine modeling error from a defensible alternative convention? Reviewers who flag every stylistic difference as a mistake are as costly as ones who miss real breaks.
  • Written precision. Your critique text is the deliverable. Vague notes like "formula seems off" don't train anything.
  • Realistic availability. Flexible does not mean absent; the lab wants consistent weekly throughput.

Logistics

Fully remote, asynchronous, your own schedule. Commitment is 5–20 hours per week with room for more if you want it. Intake includes an AI voice screen — expect follow-up questions that drill into whatever specifics you offer, so choose examples you can defend in detail. The $100/hour figure is the rate observed for this listing, not a guarantee; pay and volume can vary by task type and reviewer tier.