Research Engineer, Pretraining Scaling

About this role

Anthropic seeks a Research Engineer to own critical aspects of production pretraining pipelines, balancing deep technical work on model training systems with operational responsibilities during launches. You'll debug complex full-stack issues, optimize training efficiency, and collaborate across teams to ensure frontier models train reliably at scale.

What you'll do

• Own production pretraining pipeline including model operations, performance optimization, observability, and reliability
• Debug and resolve issues across hardware, networking, training dynamics, and evaluation infrastructure
• Design and run experiments to improve training efficiency, reduce step time, and enhance model performance
• Respond to on-call incidents during model launches with rapid diagnosis and cross-team coordination
• Build and maintain production logging, monitoring dashboards, and evaluation infrastructure
• Add new capabilities to training codebase such as long context support or novel architectures

What they're looking for

• Large-scale machine learning systems and distributed training
• JAX, TPU, PyTorch, or equivalent ML frameworks at scale
• Full-stack debugging across hardware, networking, and software layers
• Production ML systems and observability tools
• Experimental design and systems optimization
• Clear communication and cross-team collaboration
• LLM pretraining experience
• Systems engineering or operational excellence background

Benefits

• Hands-on experience with some of the largest training runs in the industry
• Work alongside world-class researchers and engineers at a mission-driven company
• Unique learning opportunities and institutional knowledge building
• 5 days per week in-office at San Francisco headquarters
• Involvement in work directly shaping safe and beneficial AI systems