Zeyuan Li

eCourtDate - Data Analyst Intern | Jan 2021 – May 2021

• Designed and optimized data architecture for Failure-to-Appear (FTA) predictions, integrating JSON API data into structured CSVs for ML workflows.
• Built interactive Tableau dashboards to visualize key FTA trends, supporting actionable insights (e.g., 10% FTA rate increase when reminders were undelivered).
• Supported feature engineering by extracting behavioral keywords from message content, improving explainability for legal stakeholders.
• Collaborated on an ML pipeline achieving 92% precision using model stacking (decision trees, random forest, gradient boosting), for a real-time court reminder system.

Ongoing: Deep Learning for Time Series Forecasting | June 2025 – Present

• Applying Recurrent Neural Networks (RNNs), including LSTMs and GRUs, to predict future trends in complex sequential datasets (e.g., stock prices).
• Focusing on handling sequential data, time-series specific preprocessing, and model optimization to enhance prediction accuracy and stability.

Ongoing: ML Model Deployment with FastAPI | June 2025 – Present

• Developing a robust REST API using FastAPI to serve a trained machine learning model, ensuring efficient model inference.
• Implementing model serialization/deserialization and leveraging Docker for containerization to create a production-ready deployment.