Financial Forecaster knowledge we require

Working the model

• Multidimensional models: navigating models, dimensions and account hierarchies, and reading a cube slice at an exact coordinate rather than at whatever grain the UI lands on.

• Scenarios: creating one, setting its parent, knowing what it inherits versus holds in its own cells, and locking it — plus identifying the forecast of record by its lock or acceptance date.

• Accounts and assumptions: telling an input account from a calculated one, and knowing when to fix a figure by editing the account formula versus overriding an assumption cell, because the two are not interchangeable and the grader can tell.

• CSV imports: loading actuals and driver data, and the unit and scale discipline an import demands.

• Note surfaces: scenario notes, driver notes, policy notes, planning notes, assumption change logs and variance commentary — and which one a given statement belongs in.

• Knowing what Financial Forecaster cannot do, so you can recognize a task that asks for a figure or a dimension member the model will never produce.

Reading models & packs

• Reading a scenario fluently: what is an input versus a calculated account, which cells are overridden, what a scenario inherits from its parent versus holds itself.

• Reading a reporting pack against the model behind it: tracing a printed figure back to the account, scenario and coordinate it came from, and telling a real tie-out from a coincidence.

• Recognizing the vintage that governs: which scenario is the forecast of record, when it was locked or accepted, and why a later forecast does not silently supersede an approved one.

• Decoding a coordinate (Time, Entity, Department, Product/Product Line, Region, and facility dimensions like Debtor or Aging Bucket) and knowing what a figure means at that grain versus aggregated.

Finding & tying out the number

• Navigating account hierarchies and scenario trees to locate the figure that actually drives a covenant, a KPI, or a disclosure — targeted retrieval instead of scrolling the model.

• Unit and scale traps: thousands-versus-dollars mismatches, a cost base that cannot be reconciled to its own revenue, per-period versus LTM versus fiscal-year-to-date, and sniffing out a figure that is off by 10× or 1,000×.

• Reconciling two in-world sources that disagree, and judging which one governs.

Entity structure & governing documents

• The reporting landscape: legal entity versus department versus segment versus consolidated, and which level a given obligation is measured at.

• Credit agreement mechanics: the provision fixing the accounting basis (frozen GAAP), Test Periods and test dates, fixed charge coverage and total net leverage, Consolidated EBITDA add-backs and their caps, Consolidated Funded Indebtedness, borrowing base and Eligible Accounts, redetermination, and who must sign a compliance certificate.

• Case relationships in the financial sense: parent and child scenarios, consolidated versus entity results, and how an amendment, consent or waiver changes the measure going forward.

Measure types & classification

• GAAP versus non-GAAP versus covenant measures, and why the same label can mean three different numbers in one pack.

• Non-GAAP discipline: exclusions requiring a written determination that an amount is not normal and recurring, reconciliation to the nearest GAAP measure, equal prominence, and restating prior periods when a metric definition changes.

• Forecast versus projection under the AICPA PFI guidance, range presentations whose endpoints must be comparably likely, and what a reasonably objective basis requires.

• The standards that decide the answer: going concern and the look-forward window (ASC 205-40), operating segments and the CODM (ASC 280), materiality screening (SAB 99 / Topic 1M), KPI disclosure (Release 33-10751), known trends in MD&A (Reg S-K Item 303), valuation allowances as a discrete item versus through the annual effective rate.

Documents, records & workflow

• Document types: compliance certificate, borrowing base certificate, lender package, board or committee pre-read, earnings release and reconciliation, disclosure committee memo, workpaper, and what each is for.

• What a finished deliverable requires versus what merely describes it: a correction stated in a memo is not the same as the correction applied in the file or written to the model.

• App workflow surfaces: scenario creation and locking, saved views, and exports for large result sets.

Note: a CPA, CFA, or MBA is welcome but not required; experienced FP&A analysts and technical accounting specialists are the primary pool.

Contract and Payment Terms

• You will be engaged as an independent contractor.
• This is a fully remote role that can be completed on your own schedule.
• Projects can be extended, shortened, or concluded early depending on needs and performance.
• Your work at Mercor will not involve access to confidential or proprietary information from any employer, client, or institution.

• Review and QA AI-agent rollouts: work through completed agent runs in Financial Forecaster and judge whether the agent's answer is correct against the actual model and documents — right scenario, right account, right coordinate, right basis, right conclusion, right deliverable.
• Identify failure modes: pinpoint where and why an agent succeeds too easily or fails wrongly, and flag tasks that are too easy or solvable without actually using Financial Forecaster.
• Improve realism and difficulty: refine the task prompt and the seeded world so tasks reflect authentic FP&A and lender-reporting work and are genuinely hard for a capable agent.
• Update grading guidance for accuracy: sharpen how answers are graded so correct answers pass and wrong ones fail, including gating failures (wrong scenario or vintage used, wrong accounting basis, an add-back taken that the agreement excludes, a figure stated but never written to the model, or a conclusion transcribed from a document that already gave it away).
• Catch plausible-but-wrong answers that a non-expert reviewer would let slide — a number that looks right at the wrong grain, or a conclusion that is correct in isolation and inconsistent with the rest of the pack.

Work arrangement: Fully remote

• Please note: We are unable to support H1-B or STEM OPT candidates at this time.

• Payments are weekly on Stripe or Wise based on services rendered.