Inspiration

Teachers and students waste hours turning study materials into quizzes manually.i wanted to make learning interactive and fast by combining AI generation with intelligent search

What it does

QuizGen-AI converts any uploaded PDF into a set of five multiple-choice questions with answers and explanations also you can enter the topic you want.It understands the document context or the topic written by the user using Elastich Search and creates smart quizzes through Gemini 2.5 Flash Note: The uploaded file must be a text-based PDF (not a scanned image), since the app extracts digital text for Elastic indexing and AI quiz generation.

How we built it

I used React + Vite for the frontend(hosted on Firebase),and a Node.js + Express backend deployed on Google Cloud Run.The backend extracts the topic ,indexes it in Elastic Serverless,retrieves the most relevant chunks,and sends them to Gemini 2.5 Flash for questions generation. All environment variables are securely managed through .env, and the app is fully cloud-based and open-source

Challenges we ran into

-Region mismatch between Elastic and Cloud Run caused DNS erros. -Handling CORS and PDF parsing issues. -Managing model fallbacks when document text was too short.

Accomplishments that we're proud of

-Fully deployed live app accessible via Firebase Hosting. -Stable end-to-end pipline : -upload→ Elastic indexing → Gemini quiz generation. -Enter a topic → Elastic indexing → Gemini quiz generation.

What we learned

-How to integrate Elastic Serverless with AI APIs. -Deploying scalable backends using Cloud Run. -The importance of error handling,API key management,and consistent prompt design.

What's next for QuizGen-AI: Elastic + Gemini-powered Quiz Generator

-Add teacher dashboards and quiz analytics. -Support multiple documents per session. -Enable exporting quizzes to Google Forms or PDF. -Inegrate real-time student performance tracking

Built With

Share this project:

Updates