EduGrade AI - World-Class NLP-Powered Classroom Evaluation Platform

EduGrade AI is a comprehensive full-stack platform designed to automate and enhance the grading process for descriptive answers using advanced Natural Language Processing (NLP) and Computer Vision (OCR).

🚀 Advanced Features

• BERT-Powered Evaluation: Uses the all-MiniLM-L6-v2 model for semantic similarity analysis between student and model answers.
• Computer Vision (OCR): Integrated EasyOCR for handwritten text extraction from photos and PyMuPDF for document extraction.
• Real-Time Data Analytics: Interactive charts using Recharts for student mastery trends and teacher performance distributions.
• Gamification & Achievements: Dynamic badge and streak system to motivate students (e.g., "Fast Finisher", "Academic Streak").
• Class Knowledge Gaps: Backend aggregation to identify frequently missed keywords, visualized for instructors.
• Modern UI/UX: Sleek, responsive glassmorphism dashboard built with React and Tailwind CSS v4.
• Secure Authentication: Custom JWT-based authentication system with role-based access control (RBAC).

🛠️ Tech Stack

• Frontend: React (Vite), Tailwind CSS v4, Lucide Icons, Recharts
• Backend (API): Node.js, Express.js, MongoDB Atlas (Mongoose)
• NLP/OCR Service: Python, FastAPI, HuggingFace Transformers, EasyOCR, PyMuPDF
• Authentication: JSON Web Tokens (JWT), Bcrypt.js

📁 Project Structure

/Evaluation
├── backend/            # Express.js server and REST API
├── frontend/           # React.js application (Vite + Tailwind)
└── nlp-service/        # Python FastAPI service for BERT evaluation & OCR

⚙️ Setup & Installation

Prerequisites

• Node.js (v18+)
• Python (v3.9+)
• MongoDB Atlas account

1. Backend Setup

cd backend
npm install
# Configure your .env file with MONGO_URI, JWT_SECRET, etc.
npm start

2. NLP Service Setup

cd nlp-service
python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`
pip install -r requirements.txt
uvicorn main:app --port 8000

3. Frontend Setup

cd frontend
npm install
# Configure VITE_API_URL in .env
npm run dev

🌐 Deployment Guide

1. Backend (Render)

• Service Type: Web Service
• Environment: Node.js
• Build Command: cd backend && npm install
• Start Command: cd backend && node server.js
• Environment Variables:
  • PORT: 5001
  • MONGO_URI: Your MongoDB Atlas URI
  • JWT_SECRET: A secure random string
  • CLIENT_URL: Your Vercel frontend URL
  • NLP_SERVICE_URL: Your Render NLP service URL

2. NLP Service (Render)

• Service Type: Web Service
• Environment: Python
• Build Command: export CARGO_HOME=/opt/render/project/src/.cargo && cd nlp-service && pip install --no-cache-dir -r requirements.txt
• Start Command: cd nlp-service && uvicorn main:app --host 0.0.0.0 --port 8000
• Environment Variables:
  • PORT: 8000
  • PYTHON_VERSION: 3.9.0 (Recommended for compatibility)

3. Frontend (Vercel)

• Build Command: npm run build
• Output Directory: dist
• Environment Variables:
  • VITE_API_URL: Your Render backend URL including /api (e.g., https://edugrade-backend-85vb.onrender.com/api)
  • CRITICAL: Do NOT omit the /api at the end of the URL, otherwise the login will fail with a 404.

📊 Evaluation & Metrics

The platform calculates:

• Semantic Similarity: Using BERT embeddings.
• Keyword Mastery: Matching extracted keywords from student responses.
• Academic Streaks: Rewarding consistent high performance.

👥 Team

• Developers: Shubashis, Vansh, Vishnu

---

© 2026 EduGrade AI. All rights reserved.