AI Training & Finance Domain Evaluation Contractor (Athyna / Confidential Client) | Remote
Reviewed finance-domain AI training and evaluation tasks for realism, data solvability, and grading consistency using expert FP&A/accounting judgment. Performed SQL-style data exploration to validate source tables, GL tie-outs, budget-vs-actuals, and AP/AR-style aging and variance logic. Assessed prompts, automated grading gates, and judgment standards to ensure tasks reflect realistic corporate finance workflows without revealing client details. • Evaluated task realism, data quality/solvability, and grading consistency • Validated tables and accounting linkages (GL tie-outs, aging, variances) • Checked prompt design and automated grading gate behavior • Ensured evaluation reflects real FP&A/accounting decisioning