Swinburne Student Project: LLM Chat Application Based on OpenRouter and Complete DevOps Pipeline Practice

This is a student course project from Swinburne University of Technology—a Python web application mimicking the Google Gemini interface, integrating large language models via OpenRouter, and implementing a four-layer DevOps pipeline from CI/CD to automated deployment. The project combines modern web development, LLM integration, and complete DevOps practices, demonstrating the automated engineering process from code submission to production deployment.

University software engineering course assignments often stay at the "just runnable" stage, which is far from production environments. As a course assignment (course code SWE40006) at Swinburne University of Technology, this project breaks away from tradition by combining modern web development, large language model integration, and a complete DevOps pipeline. It not only implements a fully functional LLM chat application but also demonstrates automated practices from code submission to deployment.

The project is a monolithic Python web application, featuring a frontend interface mimicking Google Gemini, a Python-based backend architecture, and LLM integration via OpenRouter. Its core goal is to demonstrate a complete automated DevOps pipeline. The four-layer pipeline architecture is as follows:

1. CI/CD Layer: Code hosted on GitHub, automated testing (pytest) via GitHub Actions, automatic deployment to Vercel production environment upon test pass, and preview deployment triggered by PR;
2. Monitoring Layer: Vercel Web Analytics collects visitor data and performance metrics;
3. Function Verification Layer: The application can interact with LLM APIs, supporting multi-turn conversations and context awareness;
4. Full Automation Layer: Pushes to the main branch automatically run tests, and trigger Vercel deployment upon pass without manual intervention.

1. OpenRouter: Provides a unified API interface supporting multiple LLM models, enabling flexible switching to avoid vendor lock-in, and offers free credits suitable for students;
2. Vercel: Seamless integration with GitHub, supports automatic build/deployment and preview features, global CDN acceleration, and zero-configuration deployment;
3. GitHub Actions: Natively integrated with code repositories, event-driven workflows, rich community ecosystem, and free for public repositories—ideal for student projects.

For students: Gain understanding of end-to-end development and deployment processes, exposure to modern toolchains, cultivate automation thinking, and learn LLM integration and production monitoring; For teaching design: Practice-oriented (hands-on implementation of real tech stacks), complete engineering (emphasizing testing/deployment/monitoring), and up-to-date (integrating AI and cloud-native solutions).

Reusable Patterns:

1. LLM Application Template (Gemini-style frontend + Python backend + OpenRouter + Vercel + GitHub Actions);
2. Student DevOps Checklist (Git version control, pytest automated testing, GitHub Actions CI/CD, Vercel automatic deployment, monitoring, PR preview). Improvement Directions: At the architecture level, consider microservice splitting and database addition; At the security level, need to improve authentication/authorization and API key management; At the function level, can add conversation management, file processing, and user account systems.

Although this project is small in scale, it demonstrates the trend of modern software engineering education keeping pace with the times—from "writing programs to submit assignments" to understanding the complete engineering process. By integrating LLM, CI/CD pipelines, and monitoring analysis, students experience real development workflows and cultivate engineering thinking. It provides educators with a reference template: how to combine cutting-edge technology, engineering practices, and teaching design to cultivate engineers with practical capabilities. Against the backdrop of rapid AI development, such educational practices are crucial for nurturing future software engineers.