Support Ticket Management System

A full-stack application featuring automated ticket classification using the Gemini AI API.

🧠 Design Decisions & LLM Choice

Why Gemini 2.5 Flash Lite?

• Speed: It offers the lowest latency for "on-the-fly" classification as the user types.
• JSON Mode: It reliably outputs structured data, making it easy to parse into Django models.
• Context Window: Its efficiency with short prompts makes it ideal for ticket descriptions.

Design Decisions

• Database: Used PostgreSQL over SQLite to ensure production-grade data integrity and support for concurrent connections.
• Docker Orchestration: Implemented a depends_on with a healthcheck (or sleep) to ensure the Database is ready before Django attempts migrations.
• API Optimization: Used Django's icontains for efficient ticket searching and built-in aggregation for the "Stats" dashboard to minimize frontend processing.
• Frontend UX: Implemented a "blur" trigger for AI classification to save API tokens while providing a seamless "auto-fill" experience.

🚀 How to Run

1. Create a .env file in this root directory: GEMINI_API_KEY=your_api_key_here

2. Run the command: docker-compose up --build

3. Access the App: Frontend: http://localhost:3000 Backend API: http://localhost:8000/api