Focus on developing ML models and deploying into production.

Develop, manage, and deploy machine learning solutions in the captive auto finance space, particularly supporting Residual Value initiatives. Design an analytical framework to determine the incremental value of financial incentive programs for used vehicles through the development of statistical models in combination with A/B testing principles, leading to $12M+ added value to the business.Refine and finalize developing gradient boosted decision tree models for predicting the market value of used vehicles using 15+ years of historical sales data within a Scikit-learn pipeline framework.Collaborate with machine learning engineers and software developers to deploy models into an API that allows for CI/CD functionality.Create presentations to communicate added value from data science initiatives, including presentations during company wide virtual summits.Support the business during transition from SAS to Python data science environment.

Earned the Thomas and Jean Powers Teaching Award for outstanding work as a GTA.Lead a total of five recitation classes for 150+ students for courses Introduction to Statistical Analysis (STAT 2450) and Introduction to the Practice of Statistics (STAT 1450).Created review material for topics including descriptive statistics, data visualization, probability, sampling, hypothesis testing, confidence intervals, and categorical data analysis.

Selected for a Research Experience for Undergraduates (REU) program at Carnegie Mellon University in Summer 2017 sponsored by the Center for Statistics and Applications in Forensic Evidence (CSAFE), focusing on the application of statistics to forensic science.Enrolled in lectures on probability and statistics as well as recitations to learn R programming.Constructed classification models on typing data to detect fraud by analyzing user keystroke dynamics.Designed and conducted an experiment on handwriting forgery; built models using techniques such as logistic regression and k-Nearest Neighbors to discriminate between true and fraudulent signatures.