Machine Learning

A collection of projects, particularly those utilizing machine learning and data science. See also my Github or resume.

Current Projects

• Xplore is a proprietary data platform developed for Amplity Health. With access to over 100 million medical records, I developed the machine learned components that performed information extraction and retrieval, creating rich relational data from unstructured text used for commercialization efforts, research, and health and economic outcome reporting by pharmaceutical companies.
• SanctionsExplorer is a continuously updating database and search application providing information pertaining to internationally sanctioned individuals and their known co-conspiratorsm. The unstructured, prose-styled text is spread across hundreds of goverment databases and reports from dozens of authorizing sources like the United States, US Treasury, UN, and European Union. Developed for the transnational security think tank the Center for Advanced Defense Studies
• A mapping tool of my neighborhood’s informal network of security cameras in a marginal neighborhood in Baltimore. The network was featured in the Baltimore Sun: “This Baltimore neighborhood has built an informal home camera network to fight crime. Police say it’s working.”

Learned Machines, Trained Models

• Credit Scoring the Unbanked Population in Vietnam: How do we gauge creditworthiness for new small-dollar loan applicants? I built a credit-scoring model for thin-credit history loan applicants using loan data from a bank in Vietnam (Note: dormant application; load time is a few seconds.)
• Pic-Metric (Github): Batch-user-image object detection and classification web application
• Song Recommender (Github): K-nearest neighbor model recommends thirty new Spotify songs given a user’s input of a single song based upon features such as tempo, acousticness, etc.