I am an independent AI alignment researcher based in Sapporo, Japan. I am not affiliated with a university or AI lab.

My work focuses on human-side alignment risks in non-self language models: projection, sycophancy, dependency, institutional artifact fabrication, and AI-mediated supervision failure.

I received the Cohere Labs Catalyst Grant for AI safety/alignment research and am an independent consultant through GLG Network. My single-author manuscript on self-attention and non-self cognition has passed technical check at Discover Psychology, a Springer Nature journal, and is currently with editor for editorial assessment.

I publish public research artifacts through Medium, Zenodo, GitHub, Qiita, and Zenn. My writing has appeared in AI Advances, Towards AI, and The Memoirist. My research draws on sustained long-form interaction with frontier AI systems, Buddhist contemplative practice, and long-term caregiving experience.