ClassIQ

An AI classroom platform that turns a recorded lecture into bilingual quizzes.

What it is

ClassIQ records a classroom lecture on a Raspberry Pi 5, transcribes it in Indian languages, and drafts bilingual quizzes that the teacher reviews. Students answer live or self-paced quizzes on their phones. Teachers also get timed exams, assignments with AI-suggested grading and AI teaching tools grounded in their own material. A separate super-admin console runs every school on one installation.

How it works

1. A Python agent on a Raspberry Pi 5 records the lecture in the classroom.
2. Speech-to-text for Indian languages (Sarvam AI) produces the transcript.
3. An LLM drafts bilingual quizzes from what was actually taught; the teacher reviews them before release.
4. Quizzes run on students' phones, live over Socket.IO or self-paced, as an installable web app.
5. Everything is stored in PostgreSQL; background jobs run on BullMQ; admin, teacher and student areas share one Next.js app.

Engineering and proof

• About 530 automated test cases across 54 API test files, plus Playwright end-to-end specs.
• GitHub Actions CI, Docker Compose, Caddy and Coolify deployment, backup and restore scripts.
• Multi-tenant super admin with two-factor sign-in and an encrypted secret store.
• Research notes on local models and a speech-to-text benchmark.

My role

Co-built with a team of four at BenX Solutions.

Built with

• TypeScript
• Express 5
• Socket.IO
• PostgreSQL
• Drizzle
• BullMQ
• Redis
• Next.js 16
• Python
• Raspberry Pi 5
• Docker
• Playwright

Links

Ask for a walkthrough

The code is private. A walkthrough is available on request.