Peter Cai

Built a real-time content recommendation pipeline using Python, PySpark, and Kafka, implementing partitioned event streams and fault-tolerant consumers to prevent data loss under peak traffic. Developed low-latency ML inference services with FastAPI and TensorFlow Serving and engineered feature pipelines with Airflow plus drift monitoring with Prometheus/Grafana.

Built a personalized recommendation backend using Python (Django) and Java (Spring Boot), addressing ranking/serving consistency with shared feature contracts and validation checks. Developed LLM-powered content moderation with Hugging Face Transformers and PyTorch, and designed asynchronous processing, Feast-based feature storage, model experimentation, and end-to-end observability with OpenTelem

Built financial data aggregation services using Python and Flask with normalization layers and validation rules for real-time feeds. Implemented streaming pipelines with Apache Flink and optimized PostgreSQL schemas, while creating reconciliation tools to detect and resolve discrepancies across data providers.

Built a recommendation prototype using Python and NumPy, experimenting with hybrid approaches to address cold-start issues. Designed relational data models in MySQL, created front-end visualizations with HTML/CSS/JavaScript, and improved experimental reliability with unit/integration tests.

Built React and Bootstrap interfaces to display ML experiment results for internal users and created Python scripts to preprocess datasets with missing/inconsistent data handling. Developed REST APIs with Node.js (Express), set up scikit-learn evaluation pipelines, and added automated testing with Jest and Mocha.

Completed a Master’s Degree in Computer Science at the University of California, Santa Cruz from 2019 to 2021.

Earned a Bachelor’s Degree in Computer Science at Beijing University of Posts and Telecommunications from 2015 to 2019.