Audits a financial dataset for completeness, consistency, and machine-readability. Returns a quality score, field-level breakdown, and a list of remediation actions.

Detects demographic or categorical bias in a financial dataset by comparing approval rates, average amounts, or outcome distributions across cohort groups. Returns disparity ratios and a bias risk label.

Evaluates the accuracy of AI model inferences against ground-truth labels. Computes precision, recall, F1, AUC approximation, and a pass/fail verdict against enterprise performance thresholds.

Scores each record in a financial dataset for outlier risk using Z-score and IQR methods across specified numeric fields. Returns per-record risk flags and a summary of flagged anomalies.

Compares two model variants (A and B) on the same dataset using precision, recall, F1, and AUC. Returns a winner recommendation and statistical delta.

Generates a structured KPI performance report for a financial data pipeline or AI agent run. Computes latency, throughput, error rate, and SLA adherence — formatted for senior stakeholder delivery.