Careers

About Redwood Research

At Redwood Research, we strive to ensure the safe development of artificial intelligence through rigorous technical research, thoughtful analysis, and strategic advising.

About the Role

Redwood Research staff think carefully about threat models associated with AI, carefully analyze countermeasures (through empirical methods or otherwise), and communicate their results.

Here are some example papers from our team to give an overview of the topics we work on:

• "AI Control: Improving Safety Despite Intentional Subversion"
• "Alignment faking in large language models"
• "Stress-Testing Capability Elicitation With Password-Locked Models"
• "Preventing Language Models From Hiding Their Reasoning"
• "Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats"

This role is in Berkeley, California.

Required Skills

• Machine learning. Our work involves both LLM APIs and running ML experiments on our own hardware.
• Software engineering. We frequently run into problems that require software engineering to solve; for example, orchestrating large numbers of Docker containers to run many AI agents in parallel, or running ML experiments at moderate scale efficiently.
• AI futurism. Our work is deeply influenced by our views on the future of AI, and we work better with people who are well-informed and opinionated on the topic.
• Writing. We prefer (though it's not required) that you have experience writing research papers in machine learning. It is a bonus if you have experience writing whitepapers like this.