Update 1 - Changed backend from python - 1. Supabase (Auth + Database + Serverless Functions)2. Supabase Edge Functions (token generation API)3. OpenAI Realtime API (AI backend)4. Agora Cloud (voice backend)

Update 3 - full working website built and published

Update 2 - Quiz Grading logic was not correct - resolved the issue

Update 4 - Technical Flow of Agora IntegrationIn this project, Agora is used to enable real-time, low-latency voice communication between the user and the AI tutor. The flow begins when the user initiates a voice session from the frontend. At this moment, the React application requests a secure RTC token by calling a Supabase Edge Function, which generates an Agora token using the RtcTokenBuilder with the app ID, app certificate, channel name, user UID, and an expiration timestamp. Once the token is returned, the frontend initializes the Agora Web SDK by creating a client instance via AgoraRTC.createClient({ mode: "rtc", codec: "opus" }). Using the token, the client joins the voice channel with client.join(appId, channelName, token, uid), establishing a fully authenticated and encrypted audio session. After joining, Agora automatically handles microphone capture, audio encoding, network transport, jitter buffering, and echo cancellation. The user’s microphone stream is published using client.publish(localAudioTrack), while the client listens for remote audio streams through the user-published event, subscribing to audio tracks via client.subscribe(user, "audio") and playing them with user.audioTrack.play(). This enables the platform to stream the user’s voice to the AI in real time and simultaneously play back synthesized AI voice responses generated through OpenAI's Realtime API. When the session ends, the client performs cleanup via client.leave() and stops all audio tracks, ensuring proper resource release. Overall, Agora serves as the reliable real-time audio layer, handling encoding, transmission, device access, noise suppression, and playback while seamlessly integrating with Supabase for authentication and OpenAI for conversational intelligence.