Accounting AI Research Contributor

Overview

The organization is seeking Subject Matter Experts from accounting and financial reporting (in practice or in-house) to create realistic Excel-based evaluation tasks. These tasks will test how well AI can perform work typically done by an accountant. The role involves building tasks in Excel based on your own experience and ensuring there is a clear, correct answer.

Responsibilities

• Write realistic Excel tasks drawn from your own work, including:
  • Extending existing models
  • Fixing broken ones
  • Building from a brief
• Work through each task yourself to ensure:
  • There is a clear, correct answer
• Create yes-or-no criteria that are precise enough for grading of any attempt

Requirements

• 3+ years of accounting, audit, or financial reporting under IFRS or US GAAP, in practice or in-house
• Deep, hands-on Excel skills, including:
  • Financial reporting
  • Three-statement checks
  • Purchase accounting
  • Error tracing
  • Pivot tables
• Strong written English communication skills
• Attention to detail, with willingness to learn AI and evaluation concepts

Compensation & Benefits

• Payment is task-based.
• Experts who complete their first task within the required timeline will then be qualified to do multiple tasks for the remainder of the project.
• Payments will be weekly following task completion.

Contract Details

• Engaged as an independent contractor.
• Fully remote role that can be completed on your own schedule.
• Project length: approximately 4–6 weeks.
• Start date: Immediate.
• Projects can be extended, shortened, or concluded early depending on needs and performance.
• Work will not involve access to confidential or proprietary information from any employer, client, or institution.

Task Creation & Onboarding Requirements

• Attend a 30-minute group walkthrough call.
• Create one original task from your own work and submit:
  • The brief
  • The actual input workbook(s) needed to complete it
• Input-file uploads are compulsory.
• The brief must be written by you.
• Input workbooks must be human-built.
• Use real data that has been sanitised, or public data with the source named.
• No AI usage is allowed.

How Evaluation Works

• Tasks will be run against current frontier AI models.
• If the models get the task wrong, you will write a detailed yes-or-no rubric.
• A reviewer will sign off on the task with you.

Selection Notes

• Completing onboarding does not guarantee a job offer; selection remains subject to written approval of the task proposal.

Location

Fully remote.