EduBridge AI

Ask a question — select name, grade, subject.
Auto-generated practice question with instant feedback.
AI generates a simple, grade-level explanation.
Question history tracked automatically.
Progress dashboard — accuracy and subject breakdown.

Inspiration

I'm a gap-year student from Pakistan, learning Computer Science and AI entirely through free online resources since I'm not currently enrolled in a university. This experience made me realize how much of a difference access to good learning materials makes. Millions of students around the world, especially in rural or low-resource areas, don't have access to private tutors or even reliable teachers. I wanted to build something that gives every student, regardless of income or location, a patient, always-available AI tutor.

What it does

EduBridge AI lets a student ask any question in any subject. The AI explains the concept step-by-step in simple language, adjusted to the student's grade level. After the explanation, the AI automatically generates a short practice question to test understanding, checks the student's answer, and gives instant, encouraging feedback. A progress dashboard tracks questions asked, practice accuracy, and subject-wise breakdown over time.

How I built it

I built the backend with Flask and SQLite to store questions and practice attempts. Google's Gemini API powers the explanations, practice question generation, and answer checking. The frontend is plain HTML/CSS/JavaScript, kept lightweight and low-bandwidth-friendly since the target users may have limited internet access. The dashboard uses Chart.js to visualize subject-wise progress. The app is deployed live on PythonAnywhere.

Challenges I ran into

Since I worked entirely from a mobile device, I had to adapt my usual GitHub Codespaces and terminal-based workflow to fit a smaller screen. Getting the AI to reliably return structured JSON for practice questions and answer-checking also took some prompt refinement. I also learned to handle deployment issues like WSGI configuration and Python version mismatches on PythonAnywhere.

What I learned

I learned how to design AI prompts that reliably produce both natural explanations and structured data, and how to think about accessibility and low-bandwidth design when building for under-resourced users. I also got more comfortable with full-stack deployment workflows.

What's next for EduBridge AI

I'd like to add multi-language support (starting with Urdu, since that's my native language) so it can help students who aren't comfortable learning in English, and expand the practice question types beyond short-answer format.