Tools: R (dplyr, tidymodels), Python (PySpark, Pandas, sklearn, XGBoost, LightGBM), AWS S3, CLI, Oracle SQLDeveloper, Git, GithubModeling- Trained and tuned (sklearn) a premium benchmarking model using XGBoost in Python that is used for assessing how competitively and effectively sales representatives are able to quote auto policies new customers- Trained, tuned, and maintained a household retention model using LightGBM in Python that I used to generate lifetime value estimates for current customers by combining outputs with loss ratio estimates, premiums, and financial assumptions- Trained and tuned (Optuna) a vehicle-level loss ratio model using LightGBM in Python with a Tweedie objective function that is used in combination with premium data, financial data, and retention predictions to estimate lifetime value of new Allstate customers.Data Management and Pipelining- Maintained and improved pipeline for pulling and scoring lifetime retention for large Allstate household data using PySpark and S3- Created pipeline for pulling and scoring vehicle-level loss data using PySpark and S3, leveraging different data sources for pulling and scoringInsights Generation & Delivery- Found relationship between scores of the benchmark model I built and sales; insights were delivered to sales leaders giving them proven, actionable ways to improve sales- Generated value by delivering Lifetime Value estimates by geography to optimize agency deployment- Presented findings on the lifetime value and trends of defecting customers to senior leaders- Performed initial studies on profitability of latest insurance product

- Led Agile Ceremonies, facilitated progress updates from data science team to leaders, and translated technical results to business partners- Performed roadshow presentations of a Resume Scoring Tool developed by our data science team to stakeholders- Developed a highly interactive and prescriptive Tableau dashboard used by frontline sales leaders using output data from our NLP models; equipping leaders with key metrics on sales behaviors like asking for the sale, cross-selling, and offering discounts- Developed training materials and held weekly check-in meetings with ~40 dashboard users to collect feedback and demonstrate how to use the dashboard- Performed ad hoc analyses on sales data using R - Developed a Shiny application for informing strategic growth of Allstate's independent agency presence

At Wake Forest University's Financial Services, I primarily assisted the fixed assets manager and the general accounting manager. Both teams required a substantial amount of file management from me. I would help file things physically as well as scan in documents to file electronically. For the fixed assets team, I often aided in retrieving necessary invoices from the Accounts Payable department. I also performed inventories and helped assign codes to various classrooms and locations on the Wake Forest University campus in order to help transition the department to a Workday system. For the general accounting team, I've helped key journal entries and perform bank reconciliations using both USD and international cash (Euros, Pesos, Cordobas, etc.).