Senior / Staff Software Engineer - Autonomous Agent Infrastructure

Anthropic•San Francisco, NY

1d•Hybrid

About The Position

About Anthropic 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. About the Role We're building the infrastructure that enables Claude to act in the world—booking travel, writing code, calling APIs, managing files, and completing multi-step tasks autonomously. This is foundational work for the next generation of AI capabilities. The Agent Infrastructure team designs and operates the execution environments, state management systems, and security boundaries that make autonomous AI agents safe and reliable. You'll work at the intersection of distributed systems, security engineering, and product—building systems that don't exist anywhere else in industry. This is a high-priority initiative. The problems are hard, the scope is greenfield, and the impact is immediate.

Requirements

6+ years of experience building distributed systems, infrastructure, or platform services at the hyper scale
Comfortable building Cloud Native infrastructure on GCP, AWS, or Azure
Care deeply about security, isolation, and building systems that fail safely
Have experience with containers, sandboxing, or secure execution environments (e.g., gVisor, Firecracker, V8 isolates)
Are comfortable with ambiguity—this is greenfield work, and you'll help define the architecture
Write clean, maintainable code in Python, Go, Rust, or similar
Want to work on problems that don't have existing playbooks
We require at least a Bachelor's degree in a related field or equivalent experience.

Nice To Haves

Experience building multi-tenant execution platforms or serverless infrastructure
Background in security engineering, sandboxing, or isolation technologies
Familiarity with workflow orchestration systems (Temporal, Airflow, Step Functions)
Experience with state machines, checkpointing, or durable execution patterns
Low-level systems experience (Linux internals, eBPF, container runtimes)

Responsibilities

Design and build sandboxed compute environments where Claude can safely execute code, access tools, and interact with external services
Build state management systems for long-running agent tasks—handling checkpoints, recovery, and resumption across failures
Develop authentication and authorization frameworks for delegated access—enabling Claude to act on behalf of users securely
Create observability and debugging tools for agent execution—understanding what Claude did, why, and how to make it better
Partner closely with product and research teams to define what's possible and ship it