Frontier RL Architect & Systems Engineer

Anthropic Limited; San Francisco (CA); Hybrid; USD 500,000 - 850,000; Full time

Benefits offered by this job

Equity donation matching

Vacation and parental leave

Flexible working hours

SF office space

Job summary

Anthropic is seeking an expert in reinforcement learning and large-scale AI systems to work across research and engineering on RL scaling. You will develop next-generation architectures and RL algorithms, and take them from small-scale results to frontier-scale runs, ensuring rapid iteration and reliable results across the stack, from research code to hardware.

The role focuses on building systems that maximize throughput, stability, and learning efficiency as models grow, with a location-based

Qualifications

• Experience with transformer models, training dynamics, and large-scale optimization.
• Hands-on experience training large models in distributed settings (data, tensor, and pipeline parallelism).
• Track record of original ML training or systems work (research/open-source/production impact).
• Design rigorous experiments at scale (baselines, ablations, statistical care).
• Ability to reason about compute, memory, and communication costs of models/algorithms.
• Proficient in Python and PyTorch or JAX; capable of reading/changing code at all stack layers.

Responsibilities

• Study RL training and sampling scale with model size, context length, and compute, and identify algorithmic/system changes to maintain efficiency.
• Develop next-generation model architectures and RL algorithms and run them at frontier scale.
• Translate small-scale results to frontier-scale runs and diagnose differing behavior (numerical/algorithmic/systemic).
• Build the experimental infrastructure enabling fast, reproducible comparisons of architectures and algorithm variants at meaningful scale.
• Own end-to-end performance of the largest RL runs, from research code to hardware.
• Build performance and cost models to decide which ideas should be scaled.
• Investigate training dynamics at scale, including instabilities and throughput regressions, tracing root causes.

Skills

Transformer models

Distributed training

Research track record

Experiment design

Compute cost analysis

Python & PyTorch / JAX

Tools

C++

Rust

Job description

Anthropic is seeking an expert in reinforcement learning and large-scale AI systems to work across research and engineering on RL scaling. You will develop next-generation architectures and RL algorithms, and take them from small-scale results to frontier-scale runs, ensuring rapid iteration and reliable results across the stack, from research code to hardware.

The role focuses on building systems that maximize throughput, stability, and learning efficiency as models grow, with a location-based