FraudFlux is an AI-driven card fraud detection system designed to empower B2B merchants with actionable risk intelligence. The system operates on a modern microservices architecture, featuring a Python backend (FastAPI) that processes incoming transaction payloads and enforces strict data validation schemas using Pydantic. Validated data is evaluated for risk and securely persisted in a cloud-based PostgreSQL database (Supabase). The backend will ultimately serve as the deployment environment for an XGBoost machine learning model, trained on synthetic financial data, to provide advanced predictive scoring. Finally, the system will feature a frontend merchant dashboard (React/Vite) that consumes the backend APIs, providing end-users with a live, visual ledger of transaction statuses and AI-driven risk insights.