An AI-powered assignment grading system built with Next.js (frontend) and NestJS (backend).
It allows teachers to create assignments, upload student submissions (PDFs), automatically extract and analyze text, evaluate quality using heuristics or Google Gemini AI, and export scores as CSV.

🔗 Live Demo: Auto-Grader

• ✅ Create and manage assignments (title, min word count, mode: strict/loose).
• ✅ Upload multiple student PDF submissions.
• ✅ Auto-extract student name & roll number from filename or PDF content.
• ✅ Evaluate submissions via:
  • Gemini AI (structured grading with remarks).
  • Fallback heuristic model (word count, topic relevance, structure).
• ✅ Display each student’s score & remarks.
• ✅ Export results as CSV file (downloadable).
• ✅ Responsive UI with mobile navigation (hamburger menu).
• ✅ Deployed on Vercel with runtime-safe temporary file handling (os.tmpdir()).

• Next.js 15 (Turbopack)
• TailwindCSS
• Lucide Icons
• Shadcn/ui

• NestJS
• Mongoose (MongoDB ODM)
• Multer for PDF uploads
• pdf-parse for text extraction
• csv-writer for CSV export
• Gemini API integration

git clone https://github.com/your-username/auto-grader.git
cd auto-grader