Meaify
AI-assisted handwritten exam evaluator — OCR, rubric scoring, and LLM feedback.
OCR
education
Next.js
LLM
Upload a handwritten answer sheet → vision OCR → hybrid rubric scoring → AI feedback → voice-enabled chatbot. Teacher-in-the-loop review and grading history.
Upload a handwritten answer sheet → AI-assisted OCR → rubric grading → AI feedback → voice-enabled chatbot.
Built with Next.js · TypeScript · Tailwind · Prisma + SQLite. Powered by Groq (free tier) with Gemini as failsafe and Hugging Face embeddings.
Features
- Vision OCR (Groq → Gemini fallback) on handwritten sheets
- Hybrid rubric scoring — 0.6 × semantic similarity + 0.4 × keyword match (Porter-stemmer tolerant)
- AI feedback per question, regeneratable with one click
- Voice chatbot — Web Speech API with Groq Whisper fallback
- Rubric manager — JSON rubrics under
/rubrics/, easy to git-track - Grading history — every submission persists via Prisma
Source: Utkarsh736/meaify.
Back to top