Ethan Kam

I am a Member of Technical Staff on the Integrity team at OpenAI, where I design systems for detecting and preventing abusive activity. I designed and deployed OpenAI’s first Ads Integrity systems to identify and block fraudulent impressions and clicks, combining real-time data infrastructure with machine learning models for abuse detection. I also work on autonomous investigation agents, including memory systems that help agents improve over repeated analyses, and long-running LLM workflows for evolving integrity heuristics over large-scale datasets.

My research interests are in language model behavior, fine-tuning dynamics, and data-efficient prediction methods. As an undergraduate researcher at the University of Washington, I studied fine-tuning scaling laws across model families, datasets, and parameter-efficient fine-tuning methods. This work led to Monotonic NMF, a matrix-factorization method for modeling language model fine-tuning loss curves as learned monotonic basis functions.

Before joining OpenAI, I completed my B.S. in Computer Science with a minor in Mathematics at the University of Washington (2021 - 2025). I also spent time as a quantitative trading intern on the Index Volatility group at Susquehanna International Group, where I built a model for company earnings volatility and its effect on index volatility. I previously worked as a software engineering intern at Snowflake, Palantir, and SGNL.ai, where I contributed to CAEP protocol implementation and built across data infrastructure, storage systems, and production software systems.

You can reach me at ethanwoodhill@gmail.com. My CV and selected research are available on this site.