Making AI safe at Anthropic
I'm Isa, a research engineer on the Alignment Science team. I build experiments to understand and steer the behavior of powerful AI systems.
Ask me anything about AI
A research engineer working to keep AI safe as it becomes more powerful.
I run experiments to figure out how AI systems might fail, and how to stop that from happening. Based in San Francisco.
Model organisms of misalignment
Training models to misbehave on purpose
Safety evaluations
Testing if our defenses actually work
Multi-agent experiments
AI Debate and scalable oversight
Featured Projects
Currently working on
Training model organisms to test if our safety techniques actually work
Running AI Debate experiments for scalable oversight
Building red-teaming tools to stress-test model safety
Want to chat about AI safety, alignment research, or career advice?
Always happy to connect with people interested in AI safety.