AP Duplicate Payment Detection (Client-Side)
This project demonstrates practical data analysis and audit thinking applied to real AP data problems. It focuses on explainability, conservative detection, and secure, local-only processing.
This tool identifies potential duplicate payments in Accounts Payable data using explainable, audit-grade rules. All processing runs locally in your browser — no uploads, no backend storage.
Decision-support only. This does not automate enforcement or make fraud determinations.
Why this project exists
It reflects real-world AP analysis where duplicate payments occur due to repeated invoice processing, vendor data inconsistencies, timing and rounding issues, and weak upstream controls. The design emphasizes explainability, conservative logic, and privacy-first architecture.
- Practical data analysis and audit thinking
- Rule-based detection with clear reasons
- Client-side processing for data privacy
Run a local duplicate scan
1) Upload CSV
No uploads or network calls. Your file stays in the browser.
2) Confirm column mapping
Upload a CSV file to detect headers.
3) Configure & run
Amount tolerance is fixed at ±0.01 for this version.
Results
No results yet. Upload a CSV and run the scan.
Local-only processing architecture
- CSV file read via browser APIs
- Streaming parse and detection runs inside a Web Worker
- Results stay in memory only
- No uploads or backend processing
Future enhancements (not implemented)
- XLSX ingestion via separate adapter
- Configurable rule thresholds
- Additional investigative patterns
- Optional report sharing with explicit opt-in