Software Engineer, Infrastructure, Interpretability

About the role

Anthropic is looking for a Software Engineer to lead a new infrastructure effort within its Interpretability team. You will help define the charter and build secure, private, low‑friction systems that give researchers deep access to frontier language models.

Key responsibilities

• Design, build, and own shared infrastructure for interpretability research, including secure research environments and data pipelines.
• Implement security‑by‑default architectures that enable safe model access while improving overall security posture.
• Develop privacy‑focused data‑access patterns to ensure policy compliance as research moves toward practical applications.
• Manage petabyte‑scale research data and large accelerator fleets, handling storage lifecycle, capacity planning, and scheduling.
• Create developer‑experience tooling, observability, and agentic engineering solutions that keep researchers productive.
• Collaborate closely with interpretability researchers to understand workflows and bridge communication with Anthropic’s platform and security teams.

Required profile

• Strong interest in AI safety and interpretability research.
• Ability to work closely with researchers and translate their needs into robust infrastructure.
• Experience navigating security, privacy, and large‑scale data/compute challenges.

Required skills

• Experience designing secure, private systems for high‑performance computing environments.
• Knowledge of data management at petabyte scale and accelerator fleet scheduling.
• Proficiency in building tooling and observability for developer productivity.