RL Systems Engineer - Frontier-Scale AI Performance

Anthropic Limited; New York (NY); On-site; USD 500,000 - 850,000; Full time

Job summary

The Anthropic team seeks an experienced RL specialist to bridge research and engineering work on reinforcement learning scaling. You will design scalable RL architectures and algorithms, run frontier-scale experiments, and diagnose failures across the stack from research code to hardware.

You will own performance and cost modeling for large-scale RL runs, build infrastructure for fast, reproducible comparisons, and push methods that scale while maintaining stability in AI systems.

Qualifications

• Deep familiarity with transformer language models, training dynamics, and large-scale optimization.
• Hands-on experience training large models in distributed settings, including data, tensor, and pipeline parallelism.
• Track record of original technical work in ML training or systems demonstrated through research, open-source, or production impact.
• Ability to design rigorous experiments at scale with baselines, ablations, and careful statistical treatment of compute costs.
• Ability to reason quantitatively about compute, memory, and communication costs of a model or algorithm.
• Strong programming skills in Python and JAX or PyTorch; comfortable reading/changing code at all layers.

Responsibilities

• Study how RL training and sampling scale with model size, context length, and compute; identify algorithmic and system changes for efficient scaling.
• Develop next-generation model architectures and RL algorithms; ensure frontier-scale runs run efficiently.
• Translate small-scale results to frontier-scale runs and diagnose differences due to numerical, algorithmic, or systemic causes.
• Build infrastructure for fast, reproducible comparisons of architecture and algorithm variants at meaningful scale.
• Own end-to-end performance of our largest RL runs from research code to hardware.
• Build performance and cost models for proposed architecture and algorithm changes to decide what to scale.
• Investigate training dynamics at scale, including instabilities and throughput regressions, and identify root causes.

Skills

Python

JAX

PyTorch

Experiment design

Numerical reasoning

Education

Bachelor's degree in a related field

Tools

Distributed training

High performance computing

Job description

The Anthropic team seeks an experienced RL specialist to bridge research and engineering work on reinforcement learning scaling. You will design scalable RL architectures and algorithms, run frontier-scale experiments, and diagnose failures across the stack from research code to hardware.

You will own performance and cost modeling for large-scale RL runs, build infrastructure for fast, reproducible comparisons, and push methods that scale while maintaining stability in AI systems.