About

I'm a computer scientist with a passion for research and developing AI systems that are transparent, reliable, and useful. Most recently, I spent five years as a software engineer at Google, learning how to turn research ideas into working systems.

At Google, I led a research agenda to improve open-source developer workflows with LLMs by creating a graph-structured reasoning framework to automate test-generation across AndroidOS. I collaborated with DeepMind and CoreML on data specifications for training internal LLMs, developed an unsupervised approach to root-cause analysis for Android Infrastructure errors, and led the Android Mainline API-correctness and coverage infrastructure.

Before Google, I graduated with High Honors in Computer Science from Oberlin College, with a minor in Mathematics. My honors thesis explored multi-agent reinforcement learning and game theory

I'm currently applying to PhD programs in interpretable AI, with interests in program synthesis and symbolic library learning, program-synthesized concepts for concept bottleneck models, and sample-efficient RL methods for intractably deep search problems.

As a side-passion, I make digital art with integer programming and other optimization techniques, building on independent work with Professor Robert Bosch at Oberlin. You can find the images and the methodology behind them at madebymath.art.