Viva

Inspiration

Outside of class, students often struggle to get timely, meaningful help. Emailing teachers, finding office hours, or searching online explanations doesn’t always fit into how students actually work. As generative AI becomes more accessible, it’s increasingly easy to copy answers, submit assignments, and move on — creating long-term knowledge gaps. We wanted to build something that supports learning without replacing teachers or encouraging shortcuts.


What it does

Viva is a teacher-on-demand learning assistant that helps students work independently while keeping teachers in the loop. Students upload an assignment and talk through their thinking out loud, receiving guided, Socratic feedback instead of direct answers. Teachers get visibility into where students struggle, summaries of sessions, and clear moments to intervene when needed.


How we built it

We built Viva with a modular, full-stack architecture:

  • A FastAPI backend that integrates with Google Classroom for courses, assignments, and materials
  • LiveKit to power real-time voice interaction between students and an AI tutor
  • A React + TypeScript frontend for student and teacher dashboards
  • MongoDB for persistent storage and session management
    We intentionally separated frontend and backend development so components could be built and tested independently, then integrated cleanly.

Challenges we ran into

  • Integrating multiple complex systems (LiveKit, Google Classroom, OAuth) without breaking existing functionality
  • Managing permissions, tokens, and security correctly while keeping the user experience simple
  • Designing an AI interaction that guides learning without giving away answers
  • Anticipating edge cases and bugs when connecting real-time voice with a web app

Accomplishments that we're proud of

  • A working end-to-end system that connects real-time voice, AI guidance, and classroom data
  • Secure backend integrations that keep sensitive credentials off the frontend
  • A clean separation of concerns that makes the project extensible
  • Building something that addresses a real educational problem thoughtfully, not just technically

What we learned

-How to use liveKit -how to reproduce environments in different computers

-Reading documentation for new software

What's next for Viva

  • Deeper teacher analytics and misconception detection
  • Expanded Classroom integrations (Canvas, Nexus, etc)
  • User studies to evaluate learning outcomes and iterate on the experience

Viva is just the beginning; our goal is to build tools that help students learn better and reduce friction while empowering teachers.

Built With

Share this project:

Updates