LearnMate AI

About the Project

Inspiration

The inspiration behind LearnMate AI came from a common student struggle: understanding lengthy study materials, identifying important concepts, and knowing whether we have actually learned something. Reading alone does not always guarantee understanding. We wanted to create a tool that could turn passive studying into an interactive learning experience.

What We Built

LearnMate AI is an AI-powered learning assistant that transforms user-provided study materials into simplified summaries and interactive quizzes. Students can choose their preferred summary length and quiz difficulty (Easy, Medium, or Hard). The application generates exactly five multiple-choice questions, evaluates submitted answers, calculates scores, and provides explanations for each question.

How We Built It

We developed the application using React, TypeScript, Vite, and Tailwind CSS for the frontend, with Node.js, Express, and TypeScript powering the backend. We integrated Google's Gemma model through the Gemini API to generate summaries and quizzes. We also implemented server-side validation to ensure quiz questions follow the required structure and added error handling, loading states, and API health checks for a smoother experience.

Challenges We Faced

One of our biggest challenges was ensuring that AI-generated quizzes followed a consistent structure. AI responses are not always predictable, particularly when JSON formatting is required. We addressed this by implementing schema validation and a controlled retry mechanism. Managing API rate limits, handling network errors, protecting API credentials, and keeping generation requests sequential were also important technical challenges.

What We Learned

Building LearnMate AI helped us gain practical experience in integrating large language models into real-world applications. We learned about prompt engineering, API communication, structured AI output validation, frontend-backend integration, and error handling. Most importantly, we learned that building a useful AI application requires more than connecting a model—it requires making its output reliable, accessible, and meaningful.

Our vision is simple: make learning more engaging, effective, and accessible for everyone.

Built With

• artificial-intelligence
• edtech
• express.js
• gemma
• generative-ai
• google-gemini-api
• llm
• node.js
• prompt-engineering
• react
• rest-api
• tailwind-css
• text
• typescript
• vite