Research Scientist, Interpretability - Remote

Overview

Anthropic is seeking researchers and engineers to join the Interpretability team, focusing on mechanistic interpretability of neural networks to ensure AI systems are safe and beneficial.

In Short

• Develop methods for understanding LLMs by reverse engineering algorithms.
• Design and run robust experiments in various scenarios.
• Build infrastructure for experiments and visualizing results.
• Collaborate with colleagues to communicate results.
• Strong track record in scientific research is preferred.
• Enjoy team science and collaborative discoveries.
• Comfortable with experimental science and coding.
• Ability to articulate and discuss research motivations.
• Familiarity with Python is required.
• Visa sponsorship available for qualified candidates.

Requirements

• Experience in scientific research, particularly in interpretability.
• Team-oriented mindset for collaborative work.
• Comfort with messy experimental science.
• Ability to write code and interpret results.
• Strong communication skills.

Benefits

• Competitive compensation and benefits.
• Optional equity donation matching.
• Generous vacation and parental leave.
• Flexible working hours.
• Collaborative office space.