
Case50 addresses the critical issue of "alert fatigue" in financial compliance by shifting the focus from individual false positives to high-priority risk networks. Built specifically for the Deriv hackathon, the system leverages a modern client-server architecture to achieve a target 95%+ reduction in analyst workload. Key Features & Capabilities: * AI-Powered Case Bundling: Instead of flooding analysts with individual alerts, Case50 uses graph-based algorithms (Connected Components) to cluster related entities and transactions. This reduces 2,000 raw alerts into approximately 50 comprehensive cases. * Real-Time Surveillance: The system utilizes WebSockets for sub-second scoring of live transactions, checking for specific typologies like "Rapid Fan-Out," "Structuring," and Deriv-specific risks like "Martingale" betting patterns. * Automated SAR Narratives: Integrated with Google Gemini, the platform automatically generates human-readable Suspicious Activity Report (SAR) narratives, explaining the "why" behind a case's risk score. * Regulatory Radar: A Retrieval-Augmented Generation (RAG) system allows users to query complex compliance documents (FATF, FinCEN, EBA) in natural language to assess the impact of new regulations on existing monitoring rules. * Interactive Red Teaming: Includes a "Demo Control" mode that allows users to "play God" by injecting specific money laundering scenarios (e.g., Mule Networks, Bot Abuse) to test the system's detection logic in real-time. Technology & Category Tags Technologies: * Backend: Python, FastAPI, Uvicorn, WebSockets * Frontend: React, Vite, Node.js * AI & ML: Google Gemini API, RAG (Retrieval-Augmented Generation), Graph Algorithms (Connected Components) * Data Processing: Real-time Stream Processing, Z-Score Statistical Analysis Categories: * FinTech * RegTech (Regulatory Technology) * Cybersecurity & Anti-Fraud * Artificial Intelligence / GenAI * Real-Time Data Visualization
7 Feb 2026