Chang Chen

I support incoming international students by developing resource management systems that help track mentorship activities, student progress, and engagement metrics.

I worked on an IoT coffee personalization system that analyzes saliva biomarkers to recommend optimal supplement. We onboarded 500+ beta users during testing. I developed ESP32 embedded firmware in C++ using FreeRTOS for multi-threaded sensor data collection, implementing error handling that significantly reduced data corruption. I integrated MQTT messaging with automatic reconnection to stream data reliably from devices to the cloud. On the backend, I built a Python FastAPI microservice to process incoming sensor data with connection pooling and batch writes to PostgreSQL on AWS RDS. I deployed everything on AWS EKS using Docker and Terraform, set up CI/CD pipelines through GitHub Actions, and wrote tests to ensure system reliability. The biggest challenge was building a system that bridges embedded hardware and cloud infrastructure while maintaining reliability at every layer. I learned how to design IoT systems where data integrity from sensor to database is critical, and how important testing and monitoring are when hardware and software need to work seamlessly together.

I built the backend infrastructure for a business analytics platform that uses AI to generate documents and drive user engagement. My main project was developing a real-time document generation microservice using Node.js and WebSocket, integrating the OpenAI API for token-by-token streaming so users could see content as it's being generated. I implemented RESTful APIs in Express.js for prompt handling and credit-based access control, and designed a task management system using MongoDB and Redis to support thousands of concurrent generations. I also built a serverless email automation system with AWS SQS and Lambda to re-engage trial users based on their usage patterns, which significantly improved conversion rates. The infrastructure was deployed on AWS EKS using Docker and Terraform, with CI/CD pipelines through GitHub Actions. The biggest challenge was designing a system that could scale under high concurrency while maintaining performance. I learned how to architect microservices that balance real-time functionality with operational efficiency, and how streaming responses can dramatically improve user experience and reduce abandonment.