AI Document Processor

A full-stack intelligent document processing platform powered by Azure AI Document Intelligence. Upload invoices, receipts, W-2 tax forms, and business cards — the system automatically extracts fields, line items, and structured data using AI.

Architecture

┌──────────────┐     ┌──────────────────┐     ┌─────────────────────┐
│  React 19    │────>│  .NET 10 API     │────>│  Azure AI Document  │
│  TypeScript  │<────│  SignalR          │<────│  Intelligence       │
│  Tailwind    │     │  Dapper           │     └─────────────────────┘
└──────────────┘     │  Repository       │
                     │  Pattern          │     ┌─────────────────────┐
                     │                   │────>│  Azure Blob Storage │
                     │                   │     └─────────────────────┘
                     │                   │
                     │                   │────>│  Azure SQL Database │
                     └──────────────────┘     └─────────────────────┘

Features

• Multi-Model AI Extraction — Auto-detects document type and routes to the appropriate Azure AI model (Invoice, Receipt, W-2, Business Card, General)
• Real-Time Processing — SignalR-powered live progress pipeline showing each processing step
• Document Comparison — Side-by-side field comparison with match/mismatch highlighting
• Export — Download extracted data as Excel (.xlsx) or CSV
• Approval Workflow — Review, approve, or reject processed documents with notes
• Dark Mode — Full dark theme support with system preference detection
• Dashboard — Live metrics, processing volume charts, and status distribution
• Editable Fields — Manually correct extracted values with audit tracking

Tech Stack

Layer	Technology
Frontend	React 19, TypeScript, Tailwind CSS, Recharts, SignalR
Backend	.NET 10, ASP.NET Core, Dapper, SignalR
AI	Azure AI Document Intelligence (prebuilt models)
Database	Azure SQL Database
Storage	Azure Blob Storage
Auth	Azure Entra ID (optional, JWT bearer)
Export	ClosedXML for Excel generation
Testing	xUnit, NSubstitute, FluentAssertions
Infra	Bicep (IaC), GitHub Actions CI/CD, Docker

Getting Started

Prerequisites

• .NET 10 SDK
• Node.js 20+
• Azure subscription with:
  • Azure SQL Database
  • Azure Blob Storage account
  • Azure AI Document Intelligence resource

1. Database Setup

Run the SQL migrations in order against your Azure SQL Database:

# In Azure Portal Query Editor or SSMS:
database/001-create-tables.sql
database/002-add-document-type.sql
database/003-add-approval-workflow.sql

2. Configure Secrets

cd api/DocumentProcessor.Api

# Initialize user secrets
dotnet user-secrets init

# Set your Azure secrets
dotnet user-secrets set "Azure:DocumentIntelligence:Key" "<your-key>"
dotnet user-secrets set "Azure:BlobStorage:ConnectionString" "<your-connection-string>"

Update appsettings.json with your non-secret config (SQL connection string, Doc Intelligence endpoint).

3. Run the API

cd api/DocumentProcessor.Api
dotnet run
# API runs at http://localhost:5158
# Swagger UI at http://localhost:5158/swagger

4. Run the Frontend

cd ui
npm install
npm start
# UI runs at http://localhost:3000

5. Run Tests

cd api
dotnet test
# 34 tests covering controllers and services

Docker

# Set environment variables in .env file, then:
docker-compose up --build

Deploy to Azure

Infrastructure as Code

az deployment group create \
  --resource-group <your-rg> \
  --template-file infra/main.bicep \
  --parameters infra/parameters.json \
  --parameters sqlAdminLogin=<user> sqlAdminPassword=<pass>

CI/CD

Push to main triggers the GitHub Actions deploy pipeline. Configure these secrets in your repo:

• AZURE_CLIENT_ID, AZURE_TENANT_ID, AZURE_SUBSCRIPTION_ID
• AZURE_RESOURCE_GROUP
• SQL_ADMIN_LOGIN, SQL_ADMIN_PASSWORD
• AZURE_STATIC_WEB_APPS_TOKEN

Project Structure

ai-document-processor/
├── api/
│   ├── DocumentProcessor.Api/
│   │   ├── Controllers/        # REST API endpoints
│   │   ├── Hubs/               # SignalR real-time hub
│   │   ├── Middleware/         # Exception handling
│   │   ├── Models/             # Data models & DTOs
│   │   ├── Repositories/      # Data access (Dapper)
│   │   └── Services/
│   │       ├── Extractors/    # Strategy pattern per doc type
│   │       ├── DocumentService.cs
│   │       ├── DocumentTypeDetector.cs
│   │       └── ExportService.cs
│   ├── DocumentProcessor.Tests/
│   │   ├── Controllers/
│   │   └── Services/
│   └── Dockerfile
├── ui/
│   ├── src/
│   │   ├── components/
│   │   │   ├── common/        # Shared components
│   │   │   ├── dashboard/     # Metrics, charts
│   │   │   ├── documents/     # List, detail, compare
│   │   │   └── upload/        # Upload zone, type selector
│   │   ├── hooks/             # useTheme
│   │   ├── api.ts             # Type-safe API client
│   │   └── useSignalR.ts      # Real-time hook
│   ├── Dockerfile
│   └── nginx.conf
├── database/                   # SQL migrations
├── infra/                      # Bicep IaC templates
├── .github/workflows/          # CI/CD pipelines
└── docker-compose.yml

API Endpoints

Method	Endpoint	Description
GET	/api/documents	List all documents (optional ?status= filter)
GET	/api/documents/{id}	Get document by ID
GET	/api/documents/{id}/fields	Get extracted fields
GET	/api/documents/{id}/lineitems	Get line items
GET	/api/documents/{id}/log	Get processing log
GET	/api/documents/metrics	Dashboard metrics
POST	/api/documents/upload	Upload document (multipart, optional documentType)
PUT	/api/documents/{id}/fields/{fieldId}	Update extracted field
POST	/api/documents/{id}/review	Approve/reject document
DELETE	/api/documents/{id}	Delete document
GET	/api/export/{id}/excel	Export to Excel
GET	/api/export/{id}/csv	Export to CSV
POST	/api/export/batch/excel	Batch export to Excel