How sudeeps234/gw-impactai-hackathon works
Project type and purpose
• From the file tree and README, LearnAI is a full-stack web application built in a 24-hour hackathon.
• Purpose: An accessible learning platform providing AI-powered PDF summarization, text-to-speech, and voice navigation for visually, hearing, and cognitively impaired students. There is also a separate flashcard-generation microservice that’s been prototyped.
File structure
• Top-level folders:
• client/ – React/Vite frontend
• server/ – Python Flask backend handling summarization, TTS, static audio/tts_chunks
• flashcards/ – Standalone Flask (or script) microservice for generating Q&A/flashcards
• Key files & patterns:
• client/src/components & pages → React component hierarchy (Home, Dashboard, Upload, Summary, Navbar, VoiceAssistant)
• server/app.py → main Flask API entrypoint
• server/summarizer.py, server/tts_model.py → business logic modules
• flashcards/app.py & qna_generator.py → separate service
• static assets under server/static/audio and server/tts_chunks
• No dedicated database folder—likely temporary file storage or in-memory processing.
README for additional insights
• Stack: React + Vite + Tailwind CSS on the frontend; Python Flask + AI model wrappers on the backend.
• Features: Notes summarizer, text-to-speech, voice navigation, (future) flashcards & quizzes, user dashboard planned.
• Deployment: Frontend served by Vite dev server; backend via Flask; flashcards service run separately.