Arix Gemini: The Immersive AI Tutor

A real-time AI tutor built with Gemini Live API that helps students understand complex subjects through interactive voice, image recognition, and ultra-fast performance.

Comment

Inspiration

Inspiration I’m a Grade 10 student from Sri Lanka, and I’ve always felt that education should be more personalized. With the release of Gemini 2.0, I saw an opportunity to build a tutor that doesn’t just talk but also sees and remembers. My inspiration was to create "Arix"—a companion that makes learning interactive, visual, and always available. As a solo developer, I wanted to prove that one student with a vision and the right tools can build something impactful.

What it does Arix is an immersive multimodal AI tutor. It uses Gemini Live to have natural voice conversations. It can "see" a student's work via a virtual whiteboard or screen capture. Unlike standard bots, Arix has long-term memory—it remembers your name and past lessons across sessions using browser persistence, ensuring a continuous learning journey.

How we built it I built the entire stack from the ground up:

Frontend: Next.js 14 with Framer Motion for a high-tech, responsive UI.

Backend: Python FastAPI deployed on Google Cloud Run for a scalable, production-ready environment.

AI Engine: Google Vertex AI SDK (Gemini 2.0 Flash) for real-time multimodal processing.

Pipeline: Custom-built WebSocket handlers for bi-directional audio and image streaming.

Challenges we ran into Building this entirely on my own meant I had to be the developer, the architect, and the QA engineer. Managing real-time audio synchronization was a massive hurdle; I had to solve API flooding issues by engineering a custom audio throttling pipeline. Debugging an experimental API alone taught me deep persistence and technical problem-solving.

Accomplishments that we're proud of I am incredibly proud to have built and deployed a complex, full-stack AI agent independently as a high schooler. Successfully implementing session persistence (preventing the AI from "forgetting" after a refresh) and a real-time visual-feedback loop are achievements I’m particularly excited about.

What we learned This solo journey taught me the intricacies of Google Cloud deployment, WebSocket management, and advanced prompt engineering. I realized that building cutting-edge AI is a mix of high-level architecture and gritty, low-level bug fixing.

What's next for Arix Gemini: The Immersive AI Tutor I plan to stabilize the multi-turn logic and introduce a "Live Math Solver" tool. My dream is to evolve Arix into a globally accessible platform, empowering students everywhere with a tutor that truly understands them.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Arix Gemini: The Immersive AI Tutor

Built With

  • apis-&-ai-models:-gemini-2.0-flash-(multimodal-live-api)
  • fastapi-cloud-platform:-google-cloud-platform-(gcp)-cloud-services:-google-cloud-run-(hosting)
  • framer-motion-(for-the-glowing-orb-ui)-backend-framework:-python
  • google-cloud-logging-communication:-websockets-(wss)-for-real-time-streaming-audio/video-processing:-web-audio-api
  • google-vertex-ai-sdk-frontend-framework:-next.js-14
  • html5-canvas-(for-whiteboard)-state-&-persistence:-browser-localstorage
  • management
  • react
  • session
  • typescript-styling-&-animation:-tailwind-css
Share this project:

Updates

Submission history