LLM-Powered TestGen 🚀

An intelligent, full-stack web application designed to generate dynamic aptitude (MCQ) and programming (DSA/SQL) tests with real-time, AI-powered evaluation and personalized feedback. This platform moves beyond static question banks to provide a limitless, adaptive practice environment for students and a comprehensive analytics dashboard for teachers.

---

🌟 About The Project

Traditional test preparation platforms rely on finite, static question banks, leading to repetitive practice and a limited scope for skill development. LLM-Powered TestGen solves this problem by leveraging modern Large Language Models (LLMs) to create a truly dynamic and intelligent learning ecosystem.

The application serves two primary roles:

• For Students: It offers an infinite stream of unique quiz questions for both aptitude and programming. More importantly, it provides an AI-powered agent that acts as a personal tutor, analyzing performance, evaluating code, and recommending targeted learning resources.
• For Teachers: It provides a centralized dashboard to monitor student progress, view detailed performance histories, and gain insights into the specific areas where students are struggling.

This project demonstrates a practical, end-to-end implementation of agentic AI systems in a real-world educational context.

---

✨ Key Features

• Dynamic Quiz Generation: AI-generated Aptitude (MCQ) and Programming (DSA/SQL) questions based on user-selected topics and difficulty.
• Role-Based Access Control: Secure user authentication and distinct dashboards for Students and Teachers using Firebase.
• Interactive Code Editor: A full-featured in-browser IDE powered by CodeMirror for solving DSA/SQL problems in Python, C++, and Java.
• AI Code Evaluation: An expert AI agent that provides detailed feedback on user-submitted code, analyzing its correctness, efficiency (time/space complexity), and style.
• Agentic Feedback Reports: An AI agent built with LangGraph that analyzes MCQ test results to identify weaknesses and autonomously searches for relevant video tutorials and practice materials.
• Student Performance Analytics: A personal dashboard for students to track their quiz history and visualize their accuracy trends over time with Chart.js.
• Teacher Analytics Dashboard: A comprehensive view for teachers to monitor all registered students and dive deep into their individual performance histories and AI-generated reports.

---

🛠️ Tech Stack

Category	Technologies
Frontend	HTML5, CSS3, Tailwind CSS, JavaScript, CodeMirror.js, Chart.js
Backend	Python, Flask
AI / LLM	Groq API (Llama 3.1), LangChain, LangGraph, Tavily Search
Database & Auth	Google Firebase (Authentication & Cloud Firestore)
Deployment	Render

---

⚙️ Getting Started

Follow these steps to get a local copy up and running.

Prerequisites

• Python 3.9+
• A Google Firebase project
• API keys from Groq and Tavily

Local Setup

1. Clone the repository:

   git clone https://github.com/your-username/LLM-powered-testgen.git
   cd LLM-powered-testgen

2. Create a virtual environment and activate it:

   # For Windows
   python -m venv venv
   .\venv\Scripts\activate
   
   # For macOS/Linux
   python3 -m venv venv
   source venv/bin/activate

3. Install the required packages:

   pip install -r requirements.txt

4. Set up Environment Variables:

   • Create a file named .env in the root directory.
   • Add your API keys to this file:

     GROQ_API_KEY="your_groq_api_key_here"
     TAVILY_API_KEY="your_tavily_api_key_here"
     FLASK_SECRET_KEY="a_long_random_string_for_session_security"

5. Set up Firebase Credentials:

   • From your Firebase project console, go to Project settings > Service accounts.
   • Click "Generate new private key" to download a JSON file.
   • Rename this file to firebase_admin_sdk.json and place it in the root directory of the project.
   • Important: Ensure this file is listed in your .gitignore file to keep your credentials secure.

6. Run the application:

   flask run

   The application will be available at http://127.0.0.1:5000.

---

🚀 Deployment on Render

This application is configured for easy deployment on Render.

1. Commit your code to a GitHub repository.
2. On Render, create a new "Web Service" and connect it to your repository.
3. Set the Start Command to gunicorn app:app.
4. Go to the "Environment" tab and add all the secret keys from your .env file (GROQ_API_KEY, TAVILY_API_KEY, FLASK_SECRET_KEY).
5. For Firebase credentials, add a new environment variable with the Key as FIREBASE_CREDS. For the Value, paste the entire content of your firebase_admin_sdk.json file. The application is coded to automatically use this environment variable when deployed.

---

🔮 Future Scope

• Develop an Adaptive Difficulty Engine: Implement a system where the difficulty of subsequent questions automatically adjusts based on the user's real-time performance.
• Publish a Research Paper: Conduct a formal comparative analysis of different LLMs (e.g., Llama, GPT-4, Gemini) to evaluate their effectiveness in generating high-quality test questions and pedagogical feedback, with the goal of publishing the findings.
• AI-Powered Mock Interview: Add a feature where a conversational AI acts as an interviewer, asking follow-up questions about the user's submitted code.

---

Project Link: [https://llm-powered-testgen.onrender.com/]