- Established and built the Forecasting & Supply Chain data science team responsible for all aspects of supply chain optimization, starting with probabilistic demand forecasting and flowing through to inventory purchasing and replenishment policy optimization engines. Drove cross functional forecasting and supply chain automation agenda, regularly representing strategy and progress to C-Suite- Reduced out of stock rates while reducing merchandising budget requirements by 28% annually by delivering: (1) a weekly updating probabilistic forecasting system that consistently outperformed benchmarks and expert overrides, (2) a daily inventory movement system to maintain target in stock rates and support strategic network footprint optimization, and (3) a monthly inventory purchasing system focused on efficient cash flow management- Led cross functional development of an inventory aware pricing & marketing optimization system to integrate inventory management and demand generation levers, optimizing discounted cash flow value of business by automatically triggering demand generation against excessive inventory and capturing profits where inventory is limited- Regularly collaborated with FP&A partners to deliver strategic scenario forecasts to Board of Directors and external investors, informing long term business planning and strategy and serving as company wide performance benchmark- Maintained a lean team footprint while scaling output by building a culture that prioritized candid feedback and system ownership with a high bar for written documentation and code review forums to streamline collaboration

- Scaled team from being the only forecasting scientist to a team of 8 serving as a forecasting center of excellence consulting on many use cases throughout Wayfairs operations, such as warehouse staffing, international shipping container volume, and financial risk assessment- Designed a hybrid statistical-machine learning method (Neural Network adaptation of FFORMA meta-learning approach) to forecast demand and inform all inventory decision making for Wayfair across more than 5 million products. Extended the approach into a flexible platform that was used for 11 additional use cases- Led demand forecasting efforts during COVID using executive input and qualitative research to develop scenarios to be combined with a Dynamic Factor Model leveraging case data & macroeconomic indicators, providing weekly updates and analysis to CEO & entire C-Suite. Results published in Summer 2021 issue of Foresight, the International Journal of Forecasting’s practitioner journal

- Developed genetic algorithm on top of machine learning surrogate model to optimize copper processing plant operations, increasing copper production by 30% on initial mine and 10% on additional mines including one of the largest copper mines in the world. Collaborated closely with plant operators to deploy via web toolkit that allowed for traceable collaboration between operators and system generated recommendations, work featured in the Financial Times and several external press releases- Developed a decision support tool that enabled non-technical staff to save $200M annually by simultaneously managing a portfolio of 700+ goods to control risk of stock-out, and account for substitutability, production lead times, and the relative priority of goods. Achieved this by combining state of the art forecasting techniques, stochastic optimization, and a user-friendly R Shiny dashboard- Helped a large public university improve the average SAT of their incoming class by ~45 points with no reduction in diversity or financial sustainability by developing models to understand and forecast student response to financial aid, and subsequently optimizing its allocation- Led effort to develop internal R and Python packages that encourage best practices such as model iteration tracking, data lineage management, rapid EDA, and standardized graphics templates containing 6 completed packages

Assist in data mining for research, as well as assisting students with data analysis and collection.

Used R to develop models applying both statistical and machine learning techniques to predict whole home energy usage, solar energy generation, and savings from demand response events for use by utility companies around the United States.