Designed and built machine learning and statistical models using Python (PyTorch and TensorFlow) as part of Capital One Cyber's Automated Access Control initiative.
•
Created model monitoring scripts in Python and Bash to ensure performance and reliability, optimized the models' AWS infrastructure, and built serverless functions in AWS Lambda to publish model output to consumers.
•
Led the end-to-end planning, development, and deployment of the Entitlement Risk Score statistical model, assigning quantified risk levels to Capital One's ~400,000 internal accesses at a 99.85% satisfaction rate among entitlement owners.
•
Enhanced the accuracy of a transformer-based language model for the Automated Access system by integrating a critical data source, boosting model performance by 12% for a high-impact user subset responsible for 16% of access requests.
Designed and implemented user-facing screens for the iOS (Swift) and Android (Kotlin) versions of Capital One's commercial banking app, Intellix Mobile.
•
Developed backend endpoints with Golang and contributed to the migration of infrastructure from AWS EC2 to AWS Fargate.
•
Led the development of feature toggle functionality, enabling the application to switch between states without requiring
new app releases.
•
Spearheaded the integration of the Marathon test automation tool, enhancing CI/CD pipelines by enabling automated mobile testing on emulated devices and increasing the test coverage of the Android application by 31%.
Collaborated as part of an agile software development team to enhance UnitedHealth Groupʼs Individual Insurance website, UHOne.com, aiming to improve customer retention throughout the insurance shopping experience by adding features such as application saving and help drawers.
•
Developed front-end components using AngularJS, implemented backend endpoints with .NET Core and Microsoft Azure Active Directory, and managed integrations with an Azure CosmosDB database.
Teaching a computer science course on digital logic design through the lens of the popular game, Minecraft. All concepts and techniques taught in the class are demonstrated through completing tasks within Minecraft using the in-game Redstone mechanic that mimics real-world circuitry. Main responsibilities include:
•
delivering lectures on digital logic theory and design
•
writing homework and project descriptions
•
working with and integrating class Minecraft mod with UMD's project submission server
•
creating video tutorials and other supplementary content
•
managing class logistics (grades, announcements, exams, lectures, etc)
