Research Scientist, Reasoning and Alignment

AI Breaking Wire; San Francisco (CA); Hybrid; USD 300,000 - 450,000; Full time

Benefits offered by this job

Equity packages

Health insurance

Remote-friendly culture

Parental leave

Wellness stipends

SF office spaces

Job summary

Anthropic, a safety-focused AI research company in San Francisco, seeks a Research Scientist to advance reasoning and alignment in large language models. You will design experiments, build RLHF and constitutional AI pipelines, and collaborate with engineers to scale training.

You will publish results and contribute to frontier AI understanding, while enjoying a competitive salary, equity, comprehensive benefits, and a remote-friendly office culture.

Qualifications

• Ph.D. or equivalent practical experience in Machine Learning, Computer Science, Statistics, or a related technical field.
• Demonstrated publication record in top-tier AI/ML conferences (NeurIPS, ICML, ICLR, ACL).
• Strong programming skills in Python and familiarity with PyTorch or JAX.
• Deep understanding of transformer architectures, reinforcement learning, and alignment methodologies.
• Passion for AI safety and responsible development of advanced artificial intelligence.

Responsibilities

• Design and conduct novel experiments to improve the reasoning capabilities and safety alignment of large language models.
• Develop advanced reinforcement learning from human feedback (RLHF) and constitutional AI training pipelines.
• Collaborate with infrastructure engineers to scale training runs across thousands of accelerators.
• Publish research findings and contribute to the broader scientific understanding of frontier AI models.

Skills

Python

PyTorch

RLHF

Alignment

Research

LLM

Education

Ph.D. or equivalent practical experience in Machine Learning, Computer Science, Statistics, or a related technical field

Tools

JAX

Job description

# Research Scientist, Reasoning and AlignmentAnthropic## Job Description### About AnthropicAnthropic is an AI safety and research company that's building reliable, beneficial, and interpretable AI systems. We are looking for a Research Scientist to join our Reasoning and Alignment team.### Responsibilities- Design and conduct novel experiments to improve the reasoning capabilities and safety alignment of large language models.- Develop advanced reinforcement learning from human feedback (RLHF) and constitutional AI training pipelines.- Collaborate with infrastructure engineers to scale training runs across thousands of accelerators.- Publish research findings and contribute to the broader scientific understanding of frontier AI models.### Requirements- Ph.D. or equivalent practical experience in Machine Learning, Computer Science, Statistics, or a related technical field.- Demonstrated publication record in top-tier AI/ML conferences (NeurIPS, ICML, ICLR, ACL).- Strong programming skills in Python and familiarity with PyTorch or JAX.- Deep understanding of transformer architectures, reinforcement learning, and alignment methodologies.- Passion for AI safety and responsible development of advanced artificial intelligence.### Benefits- Competitive salary with generous equity packages.- Comprehensive health, dental, and vision insurance with 100% premium coverage.- Flexible working hours and remote-friendly culture with state-of-the-art office spaces in San Francisco.- Generous parental leave and wellness stipends.## Skills & Tagspythonpytorchrlhfalignmentresearchllm## Job DetailsFull-timeSan Francisco, CA$300k – $450k USDPosted September 21, 2026Expires November 20, 2026