A selection of applied data science, machine learning, and data engineering work. Most projects include code, notebooks, and documentation on GitHub.
End-to-end ML classification framework across 958 California public high schools, evaluating 7 models under class imbalance using PR-AUC. Random Forest (PR-AUC 0.775) deployed in a live Streamlit app. Manuscript accepted for publication in the 2025 USD Capstone Chronicles.
Cloud-based predictive analytics pipeline evaluating the ROI impact of workplace health policies on productivity and absenteeism using public CDC, BLS, and County Health Rankings datasets.
End-to-end real-time data pipeline ingesting traffic, weather, and incident data from WSDOT REST APIs into a cloud-hosted MySQL database on Azure, with automated ETL scheduling and a live Tableau dashboard.
NLP-based sentiment and topic analysis comparing Reddit discussions of high- and low-performing school districts (Palo Alto vs. Oklahoma City) to surface community perception patterns.
Comparative predictive modeling of coronary heart disease risk using expanded clinical features across multiple international patient cohorts, evaluating model generalizability across populations.
Open-source Python library for reproducible EDA workflows. Features versioned releases, full pytest test suite, CI/CD via GitHub Actions, and published MkDocs documentation. Installable via pip.
A full-stack web application built with Flask and PostgreSQL, featuring user authentication, role-based authorization, and RESTful APIs. Deployed to a Linux server on AWS Lightsail.
A single-page JavaScript application using the Google Maps API and third-party data sources to deliver an interactive, map-based user experience.