Research Engineer, Pretraining Scaling

About the role

Anthropic’s ML Performance and Scaling team builds and operates the production training pipeline for its frontier language models. As a Research Engineer focused on pretraining scaling, you will ensure that large‑scale models train reliably, efficiently, and at the speed required to meet the company’s safety‑first mission.

Key responsibilities

• Own critical components of the production pre‑training pipeline, including model operations, performance optimization, observability, and reliability.
• Debug and resolve complex issues across the full stack—from hardware and networking errors to training dynamics and evaluation infrastructure.
• Design and run experiments to improve training efficiency, reduce step time, increase uptime, and enhance model performance.
• Respond to on‑call incidents during model launches, quickly diagnosing problems and coordinating solutions across teams.
• Build and maintain production logging, monitoring dashboards, and evaluation infrastructure.
• Add new capabilities to the training codebase such as long‑context support or novel architectures.
• Collaborate with teammates in San Francisco, London, and cross‑functional groups including Tokens, Architectures, and Systems.

Required profile

• Hands‑on experience training large language models or deep expertise with JAX, TPU, PyTorch, or large‑scale distributed systems.
• Enjoys a balanced mix of research and engineering work (approximately 50/50).
• Comfortable being on‑call for production systems and working long days during launches.
• Strong problem‑solving skills for ambiguous, multi‑layered issues.
• Clear communicator who can collaborate across time zones and high‑stress incidents.

Required skills

• JAX
• TPU
• PyTorch
• Large‑scale distributed systems