Research Engineer Scientist (AI)

Location: San Francisco, CA; hybrid policy requiring staff to work from one of the offices at least 25% of the time. Candidates are preferably based in the Bay Area.

Annual salary: $350,000–$500,000 USD

Company

Anthropic is a public benefit corporation building reliable, interpretable, and steerable AI systems designed to be safe and beneficial.

What you will do

• Design and run machine learning experiments for alignment science and AI safety.
• Develop techniques for scalable oversight, AI control, alignment stress-testing, automated alignment research, and model welfare assessment.
• Train language models to test the robustness of safety techniques and investigate potential alignment failures.
• Run multi-agent reinforcement learning experiments, including AI Debate-style evaluations.
• Build tools to evaluate LLM-generated jailbreaks and create scripts and prompts for testing model reasoning in safety-relevant contexts.
• Contribute ideas, figures, and writing to research papers, blog posts, talks, and responsible scaling efforts.

Requirements

• Significant software, machine learning, or research engineering experience.
• Experience contributing to empirical AI research projects.
• Some familiarity with technical AI safety research.
• Strong collaboration and communication skills in fast-moving research projects.
• Python is used for all interviews.
• Bachelor’s degree or equivalent education, training, or experience in a relevant field.

Nice to have

• Research publications in machine learning, NLP, or AI safety.
• Experience with LLMs or reinforcement learning.
• Experience with Kubernetes clusters and complex shared codebases.

Culture & Benefits

• Collaborative research environment focused on large-scale AI safety efforts.
• Frequent research discussions and cross-team collaboration with Interpretability, Fine-Tuning, and Frontier Red Team groups.
• Competitive compensation, optional equity donation matching, generous vacation and parental leave, and flexible working hours.
• Visa sponsorship is available, with reasonable efforts and immigration-lawyer support for successful candidates.
• Office space designed for collaboration.