kyuna.ai

kyuna.ai is an LLM-powered quiz application built on Python (FastAPI), Next.js and Firebase. Questions can be multiple choice, free response, or ratings, and you get put into a category at the end accordingly. While it is live, it isn't feature complete, and I'm using it as a fullstack/ml eng testbed.

Features

• Generate and serve quizzes described as human-editable YAML files
• Client app in Next.js renders quizzes, collects answers, and displays results with a modern React UI and components for quiz flow, results, and creation.
• Unique and dynamically generated shareable quiz urls
• Firebase integration for user/session persistence and optional auth flows.
• Simple CLI to load YAML quizzes into the API so content can be seeded or updated easily.
• Automated tests for both backend (pytest) and frontend (Jest + Playwright E2E) to ensure core flows and security checks remain intact.

Architecture

• Backend: Python + FastAPI. Serves JSON quiz endpoints and provides an OpenAPI surface for quick inspection. A loader CLI ingests YAML quiz files and posts them to the API.
• Frontend: Next.js (React) app that fetches quizzes and renders them. UI split into reusable components (quiz view, results view, create quiz view).
• Data/Auth: Firebase for session data and optional user management.
• Deployment: Backend runs on Render. Frontend hosted on Vercel (preview and prod envs).

Planned additions

• Image thumbnails for quizzes
• Auto-generated shareable cards for quizzes and results
• Quiz generation updates and nightly workers