
1
1
Looking for experience!

Long Description (what is your idea?) (100 words min, max 2000 characters) We built two separate AI intelligence systems for Deriv addressing distinct operational challenges. System 1: Payments Intelligence & Fraud Defense Our payments system processes withdrawal requests by computing over 50 behavioral signals: account age, trading patterns, payment history, device fingerprints, and velocity metrics. The risk engine makes millisecond decisions using Z score analysis with Median Absolute Deviation. Graph intelligence identifies coordinated fraud by analyzing shared identifiers (devices, IPs, payment tokens). When suspicious patterns emerge, the system maps fraud networks and recommends containment within 10 seconds. GenAI (Google Gemini) transforms risk signals into actionable intelligence. Every decision includes structured explanations with evidence and counterfactual analysis. Investigators use natural language queries translated into safe plans. The system auto generates incident reports and communications. System 2: Crypto Compliance Manager Our compliance platform monitors blockchain transactions across Ethereum, Polygon, and Arbitrum via Etherscan V2 APIs. Background workers poll customer wallets, detecting transactions within seconds. Anomaly detection applies statistical scoring to identify amount spikes and velocity surges. AI analysis generates compliance intelligence with risk explanations and recommendations. Regulatory Radar monitors FCA, FinCEN, FATF, and MFSA, transforming legal documents into summaries. Impact simulation tests proposed regulations against historical data. Technical Implementation Both systems use FastAPI backends, PostgreSQL for audit trails, Redis for real time features, and WebSocket notifications. React interfaces provide payment queues, fraud visualizations, crypto dashboards, and regulatory feeds. We achieve 85 to 95% auto approval rates, sub 10 second fraud detection, and comprehensive monitoring with permanent audit trails.
7 Feb 2026