Software Engineer; Sandboxing

Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.

Anthropic's sandboxing infrastructure enables Claude to safely execute code and interact with external systems. As we expand Claude's capabilities, the reliability, security, and developer experience of this infrastructure becomes increasingly critical. We're looking for an engineer to join the sandboxing team and help shape both the client-side library/API and the underlying infrastructure.

• Contribute to the client library, API surface, and underlying infrastructure for Anthropic's sandboxing system, ensuring it is reliable, well-documented, and intuitive to use.
• Drive down error rates and improve correctness through systematic debugging, monitoring, and proactive fixes.
• Help develop and maintain a product roadmap for sandboxing capabilities, balancing immediate needs with long-term architectural improvements.
• Partner closely with internal teams using the sandboxing system to understand their requirements, debug issues, and build tooling that serves their use cases.
• Respond to incidents and production issues with urgency, conducting thorough root cause analysis and implementing preventive measures.
• Build comprehensive testing, observability, and documentation to ensure the system meets a high quality bar.
• Collaborate across the sandboxing team, flexing between client-side and infrastructure work as needed.

• Have 5+ years of software engineering experience, with meaningful time spent maintaining libraries, SDKs, or developer-facing APIs.
• Obsess over developer experience—you've thought deeply about API design, error propagation, documentation, and the small details that make a library feel well-crafted.
• Have experience operating complex distributed systems.
• Bring a track record of systematically improving reliability—you've burned down error budgets, built monitoring, and driven issues to resolution.
• Can develop and articulate a long-term vision for a product, translating user feedback and technical constraints into a coherent roadmap.
• Are comfortable with ambiguity and can context-switch between reactive incident work and proactive product development.
• Communicate clearly with both technical and non-technical stakeholders.

• Experience as a founder or early engineer at an infrastructure-focused startup, where you owned a product end-to-end.
• Background in security, sandboxing, or isolation technologies (containers, VMs, seccomp, name spaces, etc.).
• Open-source contributions in the Python ecosystem.
• Experience building developer tools, CLIs, or platforms used by other engineers.
• History of working on incident response and on-call rotations for production systems.
• Exposure to reinforcement learning or model training infrastructure.

• Maintaining an open source SDK through multiple major version upgrades while minimizing breaking changes for users.
• Leading an initiative to reduce P0 incidents by XX% through improved error handling, retries, and observability.
• Building a developer platform at a startup from zero to product-market fit, iterating based on user feedback.
• Embedding with an internal team for a quarter to deeply understand their workflows and shipping targeted improvements to a piece of infrastructure they rely on.
• Developing a multi-quarter roadmap for a developer tools product, balancing user requests with technical debt reduction.

Compensation: $300,000 - $405,000 USD.

• Education requirements: At least a Bachelor’s degree in a related field or equivalent experience.
• Location-based hybrid policy: We expect staff to be in one of our offices at least 25% of the time.
• Visa sponsorship: We sponsor visas.

Your safety matters to us. To protect yourself from potential scams, remember that Anthropic recruiters only contact you from  email addresses. Legitimate Anthropic recruiters will never…