Research Scientist, Interpretability

Overview

Research Scientist, Interpretability—work on mechanistic interpretability for large language models with a focus on reverse-engineering algorithms learned in model weights.

What you'll do

• Develop methods to understand LLMs by reverse-engineering algorithms learned in their weights.
• Design and run robust experiments in toy scenarios and at scale in large models.
• Create and analyze interpretability features and circuits to understand model behavior.
• Build infrastructure for running experiments and visualizing results.
• Collaborate with colleagues to communicate results internally and publicly.

What you'll need

• Use Python for the role.
• Have a strong track record of scientific research in any field.
• Have done some work on interpretability.
• Enjoy team science and collaborative work.
• Be comfortable with messy experimental science.
• View research and engineering as two sides of the same coin.
• Clearly articulate and discuss motivations, and teach what you’ve learned.
• Write up and communicate results.

Details

• Based in San Francisco office; remote considered case-by-case for exceptional candidates.
• Location-based hybrid policy expects staff in an office at least 25% of the time.
• Minimum education: Bachelor’s degree or an equivalent combination of education, training, and/or experience.
• Required field of study: a field relevant to the role via coursework, training, or professional experience.
• Visa sponsorship is available; not guaranteed for every role/candidate.