Logo Lanfrica

kchalouah/Secure-AI-Powered-Financial-Application

Domaine:

socioeconomic

Type de record:

software
Créateur:
kch
Hôte:
This platform is a high-security, AI-driven financial ecosystem designed to protect Tunisian SMEs from fraud and financial risk while providing intelligent analytics. # Secure AI-Powered Financial Fraud Detection and Risk Management System for Tunisian SMEs with Intelligent Health Monitoring ## Full-Stack Enterprise Security Solution for Tunisian SMEs This platform is a high-security, AI-driven financial ecosystem designed to protect Tunisian SMEs from fraud and financial risk while providing intelligent analytics. --- ## 🏗 Technical Implementation Details ### 1. Secure User Management (Feature 1) - **Logic**: Implements a stateless authentication system using **Spring Security 6.5** and **JWT**. - **Code Highlights**: - `SecurityConfiguration.java`: Defines the security filter chain, enabling RBAC and stateless session management. - `JwtAuthenticationFilter.java`: Intercepts every request to validate the Bearer token and set the `SecurityContext`. - `UserServiceImpl.java`: Handles user registration with password encryption via `BCrypt` and manages account lockout logic (5 failed attempts). ### 2. Financial Transaction Management (Feature 2) - **Logic**: A robust transaction engine supporting multi-party transfers and status tracking. - **Code Highlights**: - `TransactionController.java`: Provides paginated endpoints for transaction history with role-based visibility. - `TransactionServiceImpl.java`: Orchestrates transaction creation, ensuring balance atomic updates and linking to the Fraud Detection engine. ### 3. AI-Powered Fraud Detection (Feature 3) - **Logic**: An **Ensemble Learning** strategy that runs three distinct AI frameworks in parallel to reach a consensus on fraudulent activity. - **Code Highlights**: - `FraudDetectionServiceImpl.java`: The orchestrator that calls DJL, ONNX, and TensorFlow detectors. - `DJLFraudDetector.java`: Uses **Deep Java Library** to run PyTorch/ONNX models. - `ONNXFraudDetector.java`: Directly interfaces with the **ONNX Runtime** for high-performance inference. - `TensorFlowFraudDetector.java`: Leverages **TensorFlow Java** for specialized neural network execution. - *Strategic Result* …