Used Python and state-of-the-art data mining and machine learning techniques to develop language models that predict and interpret a novel’s success based on its contents.

• Created the largest dataset (~18,000 books) for evaluating book success.
• Developed the most accurate models to date for predicting book success, with our best model achieving an average accuracy of 94.0%.
• Submitted papers to WSDM 2021 and EACL 2021 (pending decision).

Developed with C# and Python. Demonstrated that combinatorial fusion analysis can effectively predict winners and enhance play strategy of Blizzard Entertainment’s collectible card game Hearthstone.

• Combined and evaluated the performance of each combinatorial combination of five machine learning models trained on 500 Hearthstone game simulations.
• Developed a preliminary Hearthstone playing AI agent using Monte Carlo Tree Search.

Developed a system with Python that monitors loss functions of machine learning and deep learning jobs at runtime to elastically make decisions on resource configuration.

• Utilized Docker and Kubernetes for development and testing.
• Our system reduced the completion time by up to 42.06% for a single job, without sacrificing the overall makespan, in the presence of various deep learning job workloads.