AI / Full-Stack · 2025
Transaction Fraud Detection
End-to-end real-time fraud detection — Isolation Forest + JWT-secured API.
An end-to-end transaction fraud detection system combining unsupervised anomaly detection with a production-ready Flask API. Isolation Forest scores each transaction in real time, flagging anomalies based on deviation from learned normal behaviour. The Flask REST API handles authentication via JWT with role-based access control; SQLAlchemy ORM manages transaction persistence and audit logging; Pandas powers the feature engineering pipeline. An interactive dashboard visualises risk distributions and flagged transaction clusters.
Specifications
Features
Isolation Forest unsupervised anomaly detection for real-time risk scoring
Configurable risk threshold tuning for precision/recall trade-off control
Flask REST API with JWT authentication and role-based access control
SQLAlchemy ORM for transaction persistence and full audit trail
Pandas feature engineering pipeline for behavioural signal extraction
Interactive analytical dashboards for fraud pattern visualisation