AuditGuard AI：以私隱為先的本地税務及帳單審核助手

Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle Submission 💸

What I Built
AuditGuard AI is a full-stack, beginner-friendly web application designed for freelancers, small business owners, and taxpayers to automate bill auditing, category tracking, and tax compliance checks— without ever compromising sensitive financial data to third-party cloud servers.

It is built using an MVC architecture (Node.js, Express, EJS, MongoDB Atlas, and modern CSS/Vanilla JS) and features 5 core pillars:

Secure Email Verification (OTP): Real account registration protected via Nodemailer OTP verification (no console leaks, secure active user provisioning).
Smart Bill & Receipt Parser: Upload bills/receipts (PDFs or images) to automatically extract vendor names, line items, subtotals, taxes, and grand totals. 3.** Automated Audit & Math Validation**: Strict mathematical verification where Subtotal + Tax must equal Grand Total before saving or updating any invoice. Automatically flags duplicates and missing compliance data (like Tax IDs/GSTIN).
Category-Wise Analytics & Sorting: Automatically assigns bills to existing categories or creates new ones dynamically, with real-time Paid vs. Due tracking and one-click CSV report exports.
"Ask Your Bills" Local AI Chatbot: Query your invoice history in natural language (e.g., "How much tax did I pay on office supplies in September?") running locally via Ollama open-weight models.

Demo

Code

🧾 AI-Powered Audit & Tax Bill Management Web Application

Production-ready, enterprise-grade AI audit and tax invoice management portal built with Node.js, Express, EJS, Vanilla JavaScript, Chart.js, and MongoDB.

📋 Table of Contents

🌟 Overview

The AI-Powered Audit & Tax Bill Management Web Application is designed for accountants, corporate auditors, and finance officers. It automates receipt extraction, validates tax calculations (GST/VAT in Indian Rupees ₹), enforces strict mathematical integrity before saving bills, flags duplicate receipts and missing tax registration IDs, enables PII masking for sensitive numbers…

How I Built It
The application follows a clean MVC (Model-View-Controller) pattern:

Frontend: Responsive EJS templates styled with a professional, minimalist two-color base theme and clear status badges (Soft Green for Paid, Soft Amber for Due, Soft Red for Flagged/Errors).
Backend: Node.js and Express.js REST APIs secured with JWT cookies in httpOnly mode, Multer file handling, and Nodemailer transport services.
Database: MongoDB Atlas utilizing Mongoose models to store structured audit logs, user profiles, bill metadata, and dynamic categories securely.
AI & Processing: Local open-weight models powered by Ollama combined with local OCR extraction libraries to keep all financial numbers entirely offline and private.

Why Does Open Innovation Matter?
Financial data is among the most sensitive personal and corporate information in existence. Traditional tax tools rely on closed, third-party cloud APIs that require users to upload confidential tax IDs, bank numbers, and vendor details to external servers.

Open-source AI and local runtimes (like Ollama) change the game by bringing the intelligence to the data, rather than the data to the cloud. Open innovation allows developers to build trustworthy, auditable, and self-hosted financial tools that give users enterprise-grade speed combined with 100% data sovereignty.

Prize Categories

Best Use of MongoDB Atlas
Open-Source AI Integration
Side Hustle Challenge

Team Members