I am from Blacksburg, Virginia (Virginia Tech college town). I grew up going on hikes and long distance runs, which I still enjoy doing with my dog Blue. I went to Virginia Tech for undergrad double majoring in Computer Science and Applied Math, then went to the University of Michigan for my masters doing CS with a focus on AI. At Michigan I worked in labs doing embodied AI and grounded word acquisition, and helped teach classes on Affective Computing. I now work at ASAPP, a startup automating Customer Service for multiple fortune 500 companies. My current focuses are on tool use in multi turn dialogue settings, and high reasoning capabilities in low latency settings.

I am interested in improving tool calling and reasoning capabilities for agentic LLM systems. I also am very interested in investigating how to measure the soft skills of AI models, and measuring how that correlates with success in real world tasks. Finally, I have been interested on oversampling/inference scaling techniques to improve the reliability of LLM outputs.

I am currently training a reward model for ranking standup comedy sets. I am also using PPO in a custom multi turn dialogue setting to train on "ideal" synthetic dialogue data extracted using inference scaling from LLama 3.1 8B to vastly increase its tool calling capabilities at test time without any of the added latency of test time scaling.