Qnario: AI-Powered Intelligent Test Paper Generation and Online Proctoring System

Qnario is an open-source AI examination platform whose core functions include intelligent syllabus parsing, automatic test paper generation, real-time proctoring, and anti-cheating detection, demonstrating the practical application potential of AI in the education sector.

Original Author & Source:

• Maintainer: shreysoni1102-cell
• Source Platform: GitHub
• Release Time: 2025
• Original Link: https://github.com/shreysoni1102-cell/Qnario

The traditional test paper compilation process is time-consuming and repetitive. Teachers need to manually flip through textbooks to select questions and adjust difficulty, taking hours; if only part of the chapters are covered, precise question selection adds even more complexity. Qnario is designed to address this pain point: teachers upload the syllabus, AI automatically extracts units and knowledge points, and after selecting the scope, the test paper is generated instantly—reducing compilation time from hours to minutes.

1. AI Syllabus Parser

Supports formats like PDF/DOCX/JPG/PNG, automatically identifies structure to extract units and knowledge points, provides an interactive selection interface (with progress counter and select-all button), and flexibly matches teaching progress.

2. Intelligent Test Paper Generation

Dual-model strategy: The main model (Google Gemini API) generates high-quality questions, and the backup model (Groq LLaMA) switches seamlessly when Gemini's quota is exhausted; supports multiple question types (multiple choice, short answer, essay, true/false), configurable difficulty (easy/medium/hard), and teachers can preview and edit questions.

3. Real-Time Online Proctoring

• Exam Creation: Teachers generate a 6-digit room code; students join via the code, and the teacher's end displays student status in real time
• Anti-Cheating: Screen switching detection, 5-second auto-save, email OTP two-factor authentication

4. Student End & Admin Backend

Student End: Exam/practice modes, learning dashboard; Admin Backend: User management, statistics panel.

Frontend

React 18 + Vite 5, using React Router DOM 6, Axios, Socket.io-client, Recharts, jsPDF, etc.

Backend

Node.js + Express 4, paired with MongoDB + Mongoose 8, Socket.io 4, JWT, bcryptjs, Nodemailer, Multer, etc.

AI Microservices

Python + FastAPI, integrated with Google Gemini API (main) and Groq API (backup), with Pydantic v2 validation.

Database Models

Core models include 10 Mongoose models such as User, Question, StudentAttempt, StudentResult, Analytics, and GeneratedQuestion.

1. Dual-Model Fault Tolerance: Seamless switching between main and backup models to handle API quota limits and network fluctuations, ensuring user experience.
2. Scenario-Centric: Knowledge point-based question selection precisely solves teachers' pain points, reflecting an understanding of user workflows.
3. Real-Time Proctoring Technology: Socket.io for real-time communication, browser focus monitoring for anti-cheating—mature and low-cost solution.
4. Modular Architecture: AI service as an independent microservice, facilitating model switching, upgrades, and expansion.

Qnario is suitable for:

• Primary and secondary education institutions: Quickly generate weekly and monthly test papers
• Training organizations: Flexible test paper compilation based on teaching progress
• University teachers: Reduce the burden of repetitive question setting
• Online education platforms: Integration as a functional module
• Corporate training departments: Internal assessment and certification exams

Qnario is an open-source project that deeply integrates AI with education scenarios. It focuses on the engineering implementation of the complete process from 'syllabus parsing → intelligent test paper generation → online exam → real-time proctoring'—not pursuing large model parameter scales but emphasizing practical落地 (practical implementation). It has reference value for developers working on AI education applications and educators looking for test paper systems.