Accounting Expert (QuickBooks) — AI Model Evaluation

Contract · Remote · $120/hr

Role Overview

Huzzle is seeking experienced accountants to test how well frontier AI systems handle real accounting work. You will write realistic accounting prompts that are difficult enough to make a frontier model fail, and then write a rubric for each one that lets any reviewer grade the model's output consistently. A rubric is a list of objective, checkable criteria — a reviewer applies it to an answer and marks each criterion pass or fail. Training is provided. Prior AI or data-annotation experience is preferred.

Key Responsibilities

• Write realistic accounting tasks of the kind a client, controller, or business owner would send
• Month-end close, bank and credit card reconciliations, AR/AP ageing, accruals and adjusting entries, sales tax, job costing, and cleaning up a messy QuickBooks file
• Each request should call for a working deliverable, usually a spreadsheet, a journal entry schedule, or a short memo
• Run the request against a frontier AI model and examine the output with the scrutiny you would apply to work you were signing off. Where the model succeeds, escalate the difficulty until a genuine gap appears
• Author a scoring rubric for each task with objective criteria another reviewer could apply without repeating your work, for example, "the reconciled ending balance equals $X" or "the $X deposit is recorded as a customer deposit liability, not revenue."