Transaction anomaly visualizer
Distributed transaction anomaly detection using Neo4j graph traversal, BullMQ job queues, and a React dashboard — open-sourced as a standalone Node.js package.
Transaction Anomaly Visualizer (TAV)
Detects fraudulent patterns in financial transaction networks by modelling accounts and fund transfers as a property graph in Neo4j.
⚙️ CORE CAPABILITIES
🔁 DFS cycle detection — circular fund flow identification
⚡ BFS velocity checking — anomalously rapid successive transfers
📊 Threshold proximity analysis — amount-based flagging
⏱️ Timestamp delta computation — time-pattern anomalies
🏗️ ARCHITECTURE
🗄️ PostgreSQL — raw transaction storage
🕸️ Neo4j — property graph model & traversal
📨 BullMQ + Redis — async distributed job queue
🌐 REST API — anomaly exposure layer
⚛️ React dashboard — flagged subgraphs & traversal paths
📦 OPEN SOURCE
Core detection engine ships as a standalone Node.js npm package — fully decoupled from the UI and server.