GradeOps

GradeOps is an AI-powered evaluation infrastructure designed to bridge the gap between handwritten student submissions and digital academic grading. It automates transcription (OCR), applies instructor-defined rubrics via AI, and provides a secure dashboard for TAs and Professors.

---

Demo

Check out GradeOps in action:
🎥 Watch the Demo on Google Drive (!! Only frontend is deployed on vercel)

---

Features

Instructor Hub

• Bulk Upload: Upload scanned student answer sheets (JPG/PNG/PDF).
• Dynamic Rubrics: Define custom grading criteria and point weightage.
• Evaluation History: Manage and track past grading records.
• TA Handover: Seamlessly transition results to the TA Dashboard for review.

---

TA Dashboard & Reviewer

• AI Scoring: View AI-generated scores and feedback for each question.
• Integrity Check: Automatic plagiarism and similarity detection flags.
• Live Overrides: Teaching Assistants can manually adjust marks with real-time grade recalculation.
• CSV Export: Export final results for official record-keeping.

---

Security

• Gatekeeper: Password-protected access to return to the Instructor Hub.
• Environment Safety: Sensitive API keys and backend configurations are handled via .env (excluded from tracking).

---

The AI Pipeline

GradeOps uses a sophisticated multi-stage pipeline to transform physical handwriting into actionable academic insights.

---

1. The Vision Layer (OCR)

The system ingests raw image data (JPG/PNG/PDF) and processes it through an Optical Character Recognition (OCR) engine.

• Digitization: Converts handwritten strokes into digital text.
• Structural Mapping: Identifies specific answer blocks corresponding to the exam layout.

---

2. The Intelligence Layer (Mistral AI)

Once digitized, the text is fed into the Mistral AI model (via API) alongside the instructor's custom rubric.

• Semantic Evaluation: Mistral analyzes the meaning and context of the student's response rather than relying only on keywords.
• Rubric Alignment: The AI measures responses against the instructor-defined grading criteria and weightage.
• Feedback Generation: Generates human-like qualitative feedback explaining the assigned score.

---

3. The Integrity Layer

Simultaneously, the content is analyzed for:

• Plagiarism Detection: Cross-referencing submissions to identify high similarity scores.
• Consistency Checks: Ensuring OCR transcription quality and academic coherence.

---

Tech Stack & Integration

• Frontend: React.js
• Backend: Python (FastAPI)
• OCR Engine: Tesseract OCR / EasyOCR
• LLM: Mistral AI (Mistral-Large / Mistral-Medium)
• Database: MongoDB
• Storage: Cloudinary
• State Management: React Hooks & Local Overrides
• Styling: Premium Bento-grid UI with Dark Emerald aesthetic

---

Installation & Setup

1. Clone the Repository

git clone https://github.com/sargamjain5/gradeOps.git
cd gradeOps

---

2. Frontend Setup

cd frontend
npm install
npm start

---

3. Backend Setup

cd backend
pip install -r requirements.txt
python main.py

---

4. Create a .env File

Create a .env file inside the backend/ directory and add:

MISTRAL_API_KEY=your_mistral_api_key

MONGO_URI=mongodb://localhost:27017/gradeops

CLOUDINARY_CLOUD_NAME=your_cloud_name
CLOUDINARY_API_KEY=your_cloudinary_api_key
CLOUDINARY_API_SECRET=your_cloudinary_api_secret

---

Default Credentials

Instructor Login

Username: instructor
Password: 1234

---

TA Login

Username: ta
Password: 1234

---

Instructor Dashboard Access Password

admin123

---

License

This project is intended for educational and research purposes.