Lead Multi-Agent RL Research Scientist

Anthropic Limited; Seattle (WA); On-site; USD 500,000 - 850,000; Full time

Benefits offered by this job

Competitive compensation and benefits

Equity donation matching (optional)

Generous vacation & parental leave

Flexible working hours

Office space in Seattle

Job summary

Anthropic is seeking researchers to advance training of multi-agent systems at scale and to refine reinforcement learning for language models. You will design environments, experiment with agent harness configurations, and develop rigorous benchmarks while collaborating with product teams to apply agents in real-world AI applications.

The role emphasizes building safe, interpretable, steerable AI and contributing to a mission-driven organization with a strong emphasis on collaboration and impact.

Qualifications

• Bachelor's degree or equivalent required.
• Experience with large-scale RL on language models and multi-agent systems.
• Excellent communication and collaboration with researchers on difficult tasks.
• Passion for safe, societally beneficial AI and mission-driven work.

Responsibilities

• Help design and optimize environments and data for model training that maximize Claude’s performance and ease of use with agentic tasks.
• Ideate and compare agent harness configurations (memory, context management, communication architectures).
• Design and implement rigorous quantitative benchmarks for large-scale agentic tasks.
• Collaborate with product teams to apply agents effectively in products.

Skills

Large-scale RL

Multi-agent systems

Communication skills

Safety/AI ethics

Collaborative research

Education

Bachelor's degree

Job description

Anthropic is seeking researchers to advance training of multi-agent systems at scale and to refine reinforcement learning for language models. You will design environments, experiment with agent harness configurations, and develop rigorous benchmarks while collaborating with product teams to apply agents in real-world AI applications.

The role emphasizes building safe, interpretable, steerable AI and contributing to a mission-driven organization with a strong emphasis on collaboration and impact.