I worked under Prof. Heewook Lee on a HMM online planning strategy for blackjack written in Python, and in the CACTUS Lab under Prof. Jia Zou on extending a queryable database neural network using Python and Tensorflow.
Developed and deployed a full-stack fraud assistance application to production, increasing resolution efficiency by 63% for hundreds of customers. Also built a serverless API and related microservices, as well as hardening the underlying infrastructure using RSA encryption and OAuth authentication to support application scaling.
Developed significant portions of the company’s flagship Dashboard passwordless identity platform using Ruby on Rails, and wrote Ruby modules for an integrations state machine alongside services for customer event alerts and QR response tooling. Wrote a suite of 20+ stateless end-to-end consumer authentication integrations using Python, Java, and Ruby for third-party partner API usage. Also leveraged Ruby to create a multi-threaded pooled concurrency library, which optimized native process runtime by 35%.
Worked in the Data Science Group under Prof. Elke Rundensteiner on an enhanced positive-unlabeled (PU) gradient boosting classifier in PyTorch. The model beats standard methods in the scikit-learn Python library by 8.3% on average.
