Principal Investigator

Delivered a machine-learning model that identifies online shoppers most likely to purchase a vehicle in the near term.

Sales teams were investing valuable time in leads that rarely converted, so by analyzing historic CRM and web-analytics data we produced a propensity model that highlights genuine purchase intent. Qualified lead volume rose 22 percent and average deal closure occurred five days sooner, with sales managers providing feedback that informed ongoing tuning. The solution used pandas for feature engineering, XGBoost deployed behind a FastAPI microservice, and a Tableau dashboard for transparency.

Prototyped an interactive voice bot that helps nurses complete routine patient check-ins more efficiently.

Home-care nurses were spending significant time on routine calls, so in collaboration with clinical stakeholders we designed a conversational assistant that captures key health indicators and escalates when required. A pilot showed a 22 percent reduction in call duration while maintaining positive patient feedback, and nurses appreciated reclaiming time for direct care. The prototype combined Dialogflow, TensorFlow TTS, and Twilio Voice, with iterative usability testing guiding refinements.