Research Engineer / Research Scientist, RL Frontiers (AI)

Location: San Francisco, New York City, or Seattle, United States; staff are expected to work from one of the offices at least 25% of the time.

Annual salary: $500,000–$850,000 USD

Company

Anthropic is a public benefit corporation developing reliable, interpretable, and steerable AI systems.

What you will do

• Study how reinforcement learning training and sampling scale with model size, context length, and compute.
• Develop next-generation transformer architectures and RL algorithms and run them efficiently at frontier scale.
• Scale small-model results to large training runs and diagnose numerical, algorithmic, and systems-level differences.
• Build experimental infrastructure for fast, reproducible comparisons of architecture and algorithm variants.
• Own end-to-end performance of large RL runs, from research code to accelerator hardware.
• Investigate instabilities, divergence, throughput regressions, and compute and cost trade-offs.

Requirements

• Deep familiarity with transformer language models, their architectures, training dynamics, and large-scale optimization.
• Hands-on experience training large models in distributed settings, including data, tensor, and pipeline parallelism.
• Original technical work in ML training or systems demonstrated through research, open source, or production impact.
• Ability to design rigorous large-scale experiments with baselines, ablations, and statistical validation.
• Quantitative understanding of compute, memory, and communication costs.
• Strong programming skills in Python and JAX or PyTorch, with the ability to modify code across the stack. A bachelor’s degree or equivalent experience is required.

Nice to have

• Research experience in reinforcement learning, optimization, or large-scale training.
• Experience developing RL algorithms for language models, scaling laws, or quantitative training-efficiency models.
• Experience modifying transformer architectures and scaling training across large accelerator fleets.
• Understanding of low-precision numerics, training instability, and GPU or TPU performance characteristics.
• Experience with C++ or Rust.

Culture & Benefits

• Collaborative research environment focused on large-scale efforts in safe and steerable AI.
• Frequent research discussions and emphasis on communication and empirical science.
• Flexible working hours and office-based collaboration.
• Competitive compensation, optional equity donation matching, generous vacation, and parental leave.
• Visa sponsorship may be available, with immigration-lawyer support, depending on the role and candidate.

Hiring process

• Applications are evaluated against role-specific qualifications and internal job-level requirements.
• Candidate AI usage is governed by Anthropic’s application policy.