- Identified millions in contracting opportunities by building a machine learning system classifying genetic testing claims into comparable tests- Operationalized the machine learning system so that other users could easily add training set examples, retrain models, and try different models using a RESTful API- Containerized the machine learning system to be used in different cloud platforms (e.g. AWS Sagemaker)- Contributed to sales efforts by performing what-if scenarios on claims data for different tiers of service- Built prototype claim editing API and used it to identify millions in policy violations for large health insurers- Systematized project management by championing agile methods which were ultimately adopted by the data science team

Consulted with clients on applications for influencing social networks to facilitate business development of emerging countries and detecting anomalous actors within social networks, resulting in published work

- Reduced downside risk with potential multi-million savings by creating a hierarchical modeling system for forecasting energy load in the day ahead market.- Eliminated discrepancy between price and valuation of assets caused by different pricing and valuation apps by inserting calls to newer pricing app within legacy pricing system. Additional impacts included: obviation of retraining through a seamless, unchanged user experience and reduced project completion time by reusing swaths of existing code within the legacy pricing app.- To increase throughput of Monte Carlo valuation and pricing of assets with faster scenario analyses, architected a distributed simulation system.- Sped up asset pricing workflows by working as the product owner of agile team in creating, detailing, and prioritizing stories for a team of seven IT professionals.- Reduced valuation errors and created more transparent checks and balances by automating position reports and building several command line tools for extracting and identifying new warnings for the valuation app.- Reduced space and speed up valuation simulations by 19% by researching optimal compression of correlation matrices over the space of low-rank correlation matrices as applied to model calibration artifacts.- Removed barriers for programmatic use of the valuation app by writing wrapper code around valuation app's RESTful API.

- Taught class on scientific computing- Conducted research on model-based semi-supervised clustering problems and inference on doubly stochastic messaging activities- Performed numerical experiments, produced figures, proved theorems, and co-authored papers

Proposed and implemented integration of an improved optimization procedure forthe main product, which used Map-Reduce principles to perform concurrent calculationson large-scale communications data to model social behaviors

- Increased annualized state policy growth by 30% as measured by comparing performance month before to month after change by filing changes to a state's merit plan that used an analysis of loss experience by points based on accident and legal violations. - Increased company growth while maintaining profitability by pricing and filing targeted changes to private passenger automobile insurance rates for several states.- Realigned geographical risk in a state by using ARC-GIS software to analyze and refine territorial definitions.- Streamlined rate review process by acting as the project owner of the department's most commonly used Visual Basic pricing program.- Developed colleagues' technical skills by teaching a Visual Basic for Applications class for the departmental staff development program.- Greatly sped up the time to perform an increased limit factor analysis by creating a VBA/SAS program to generate exhibits.