I'm Zephyr Cady — an engineer and AI practitioner with 13 years of experience across cybersecurity, project management, and high-assurance systems. I currently lead a data science and AI team at the Naval Nuclear Laboratory, where I oversee applied ML and AI initiatives supporting the U.S. Navy's nuclear propulsion program. This site is where I think out loud about AI safety — technical alignment research, eval methodology, and what rigorous engineering culture from high-consequence domains has to offer the field. I'm particularly interested in the intersection of control, interpretability, and deployment safety in critical environments. If you're working on evals, red-teaming, or alignment research and want to talk, reach me at zephyrcady@gmail[.]com.