Live URL: https://form-ai-rosy.vercel.app/
- username: test
- test@123
Demo Video: https://www.loom.com/share/8883609e648e47158b61e5ce44612409?sid=0a8d11dd-6caa-4dcc-a182-a03ce94ac3fa
- PDF file upload and storage
- Automated text extraction from PDFs
- AI-powered text analysis using Gemini 1.5 Flash
- Secure authentication system
- RESTful API architecture
- Runtime: Node.js
- Framework: Express.js
- AI Model: Google Generative AI (Gemini 1.5 Flash)
- Database ORM: Prisma
- Storage: Supabase Bucket
- Authentication: JWT (JSON Web Tokens)
- Validation: Zod
- File Processing: pdf-parse, multer
The application is built using a microservices-style architecture with three main services:
- Handles PDF file uploads through multer's memory storage
- Securely stores files in Supabase bucket
- Returns public URL and file metadata
- Implements robust error handling and validation
- Receives PDF public URL from Upload Service
- Extracts text content from PDF documents
- Utilizes pdf-parse for reliable text extraction
- Processes documents maintaining original formatting
- Processes extracted text using Gemini AI
- Generates dynamic labels and values based on document content
- Provides intelligent text analysis and classification
- Optimized for accuracy and performance
The service implements secure authentication using JWT:
- Required for protected routes
- Token-based authorization using Bearer scheme
- Middleware protection for sensitive endpoints
Base URL: /api/v1
| Endpoint | Method | Authentication | Description |
|---|---|---|---|
/ |
GET | No | Home route |
/signup |
POST | Yes | User registration |
/signin |
POST | Yes | User authentication |
/upload |
POST | Yes | PDF upload endpoint |
/nlp/:dataId |
GET | Yes | AI analysis endpoint |
The service is deployed on Render
- All routes handling sensitive operations are protected with JWT authentication
- File uploads are validated and sanitized
- Secure storage implementation using Supabase
- Input validation using Zod schemas