Training Turk

Mercor listing

FP&A Expert

$70–110/hr

The role in one line

An FP&A or technical accounting expert who audits AI agent performance in Financial Forecaster against real models, covenants, and reporting obligations.

Written by Training Turk from the public listing; it may be incomplete or out of date. Read the full posting on Mercor.

What you would do

  • Review AI agent runs in Financial Forecaster and verify answers against actual models and source documents
  • Identify when agents use wrong scenarios, vintages, accounts, or accounting bases that produce technically wrong answers
  • Catch plausible-but-wrong conclusions that non-expert reviewers would accept as correct
  • Refine task scenarios and grading criteria so FP&A evaluation reflects realistic practitioner work
  • Flag when tasks are solvable without actually using the model or when model mechanics are unclear

Who they are looking for

  • 5+ years in financial planning, accounting analysis, financial reporting, technical accounting, controller functions, or lender-reporting specialist roles
  • Weekly or more hands-on use of Financial Forecaster for real forecasting, covenant, or reporting work
  • Ability to read models fluently, identify inputs versus calculations, spot overrides, and judge figure accuracy at stated grain
  • Deep familiarity with credit agreement mechanics, Test Periods, add-backs, leverage ratios, and borrowing base requirements
  • Sound judgment on accounting basis, GAAP versus non-GAAP discipline, and going concern or segment reporting frameworks

Skills this role asks for

financial forecaster model navigationscenario and vintage identificationmultidimensional modeling expertisecovenant compliance verificationcredit agreement mechanicsunit and scale disciplineaccount hierarchy interpretationoverride and input detectiongaap and non-gaap reportingborrowing base and leverage calculation

What the interview is likely to probe

  1. 1.Scenario and vintage identification in complex models

    Evaluators check whether you instantly recognize the governing scenario, its lock or acceptance date, and which prior period the forecast inherits, catching timestamped data errors.

    Expect something like: “An AI agent says borrowing base is 62 million from Q3 forecast. Q4 forecast revised it to 68 million. Would the Q3 answer be marked right or wrong?”

  2. 2.Override detection and account hierarchy navigation

    Interviewers test whether you spot when agents confuse input accounts with calculated ones, miss cell overrides, or read aggregated figures at the wrong hierarchy level.

    Expect something like: “An agent claims Consolidated EBITDA is 87 million, but that cell is overridden. The formula would produce 92 million. How would you evaluate whether the agent found the right figure?”

  3. 3.Unit and scale trap identification

    Evaluators verify you catch when AI converts thousands to dollars incorrectly, mixes LTM with fiscal-year-to-date, or reads a figure off by 10x or 1,000x due to scaling.

    Expect something like: “An AI calculates leverage as 4.2x by dividing Indebtedness in millions by EBITDA in thousands. Would you flag this calculation error and what would the correct ratio be?”

  4. 4.Credit agreement mechanics and add-back discipline

    Interviewers probe whether you understand add-back restrictions, Consolidated EBITDA treatment, redetermination clauses, and which metrics are measured at consolidated versus entity level.

    Expect something like: “The credit agreement caps Consolidated EBITDA add-backs at 15% of baseline. An AI proposes severance of 8 million and acquisition expenses of 3 million. Would both be eligible?”

  5. 5.Reporting tie-out and document reconciliation

    Evaluators test whether you trace printed figures through the model and documents, catching conclusions that are correct in isolation but contradict other parts of the pack.

    Expect something like: “An earnings release states Operating Cash Flow increased 12% year-over-year. A board pre-read shows a 3 million working capital swing the model doesn't capture. Is the comparison faithful?”

Exercise you may get

Review a complete AI run on a covenant compliance certification task, including task prompt, agent's reasoning, model outputs, and credit agreement language, then determine whether the answer is correct, salvageable, or fundamentally wrong.

How to prepare

  • Review recent covenant certifications you have signed or supported to internalize the approval standards
  • Study the credit agreement governing your organization's or a peer's borrowing to understand add-back restrictions and measurement bases
  • Practice tracing a complex figure from an earnings release back through the model and source documents to identify discrepancies
  • Document two scenarios where AI or less-expert colleagues misidentified the governing forecast or applied the wrong accounting basis

Facts

Pay
$70–110/hr
Commitment
hourly
Hours
40 per week
Work arrangement
remote · Remote
Company
Deeptune Internal
Posted
9/18/2026