research, engineering, and technical notes
I’m an AI engineer and independent AI safety researcher working on LLM agents, evaluations, strategic deception and dynamic subspace regulation in transformer models. I’m interested in how internal transformer components regulate directions to balance competing behavioral objectives, and how these mechanisms emerge during training.
Previously, I was a data scientist in applied NLP and machine learning, spanning projects from one-off analyses and applied research to production AI systems.
Outside of my work, I’m probably traveling, making music or occasionally tossing around yo-yos.