Pre-training Distributed Systems Tech Lead / Manager

About the role

Anthropic is building safe, interpretable AI and needs a senior technical leader for its Evaluation Infrastructure team. The role focuses on designing, scaling, and operating the distributed systems that run large‑scale model evaluations, ensuring that measurement of model capabilities is fast, reliable, and trustworthy.

Key responsibilities

• Lead the team that builds the distributed scheduler, orchestrator, and executor for frontier model evaluations.
• Own evaluation throughput and cost, managing compute allocation, queueing across accelerator pools, and caching of evaluation work.
• Develop and scale the harnesses researchers use to define, run, and iterate on evaluation suites.
• Guarantee determinism, reproducibility, and honest uncertainty quantification of evaluation metrics.
• Deliver evaluation signals to dashboards and review processes that inform launch decisions.
• Contribute as an individual contributor while hiring, coaching, and prioritizing the team’s work.

Required profile

• Proven experience leading end‑to‑end projects on large‑scale distributed systems.
• At least one year of experience managing engineers or serving as a tech‑lead with direct reports.
• Strong Python and Rust programming skills.
• Track record building high‑throughput, fault‑tolerant systems on cloud or on‑prem accelerator fleets.
• Passion for measurement quality and ability to detect metric drift for the wrong reasons.
• Excellent communication with researchers and ability to translate research needs into infrastructure.
• Deep interest in safe, transformative AI development.

Required skills

• Python
• Rust
• Distributed systems design
• High‑throughput and fault‑tolerant system engineering
• Cloud or on‑prem accelerator fleet management
• Observability and monitoring
• Statistical analysis for evaluation metrics