eazytbay/Veridoc_ng

AI-powered invoice and receipt verification system for fraud prevention in Nigeria, built with the MERN stack. # VeriDoc NG **AI-Assisted Invoice and Receipt Risk Verification System for Fraud Prevention in Nigeria** VeriDoc NG is a deployable MERN stack SaaS project designed for Nigerian SMEs, finance teams, schools, NGOs, cooperative groups, fintech operations, and compliance analysts who need a safer workflow before approving payments or releasing funds based on invoices or receipts. The application does **not** claim to prove an invoice is genuine with 100% certainty. It performs **risk-based verification** by extracting document information, applying fraud-control rules, comparing against approved vendor records, identifying duplicates and suspicious patterns, and generating a clear analyst-friendly risk explanation. ## What is included - MERN architecture: MongoDB, Express, React, Node.js - Authentication with JWT - Role-based access: Admin, Supervisor, Analyst - Client-facing landing page - Modern responsive dashboard UI - Invoice/receipt upload flow - OCR/text extraction service using Tesseract.js and PDF text extraction using pdf-parse - Rule-based verification engine - Risk scoring: Low, Medium, High - AI risk explanation hook with OpenAI support and local fallback summary - Vendor/account matching - Duplicate invoice detection - Missing field checks - Suspicious date checks - Similar invoice checks - Verification history - Supervisor review: approve, reject, escalate - Audit trail logging - Subscription plans and document usage limits - Demo plan upgrade flow - Seed data for testing - Deployment-ready environment files ## Project structure ```txt veridoc-ng/ backend/ Express API, MongoDB models, OCR, verification engine client/ React + Vite + Tailwind frontend docker-compose.yml README.md ``` ## Tech stack ### Frontend - React - Vite - Tailwind CSS - React Router - Axios - Lucide React - Recharts ### Backend - Node.js - Express - MongoDB + Mongoose - JWT authentication - Multer file upload - Tesseract.js OCR - pdf-parse - OpenAI SDK optional …