Software Engineer, Sandboxing

Details

Anthropic's Sandboxing team builds the isolation layer for code execution, computer use, and agent workloads. The systems run untrusted, model-generated code and support training, evaluation, and customer-deployed agents. The role operates at the boundary of the Linux kernel, virtualization, and container ecosystems, with a focus on secure, fast, and highly scalable sandbox execution.

Responsibilities

• Design, build, and operate sandboxing infrastructure for untrusted code and agent workloads at scale.
• Improve sandbox startup time, density, and throughput through snapshot and restore, checkpointing, and warm pools.
• Strengthen isolation across compute, filesystem, and network, including egress controls and per-session credential handling.
• Investigate and resolve performance, reliability, and security issues at the kernel, hypervisor, and container-runtime levels.
• Build monitoring, tracing, and debugging tools to analyze time and resource usage inside sandboxes.
• Support reinforcement-learning environments and evaluations requiring fast, reproducible, isolated execution.
• Partner with product and platform engineers to expose sandboxes as dependable capabilities for Claude products and customers.
• Threat-model the platform and test, harden, and respond to escape and abuse scenarios with security teams.
• Work with cloud providers on required hardware and platform features.
• Share knowledge of systems programming and Linux internals and help establish technical standards.

Requirements

• Software engineering experience in Linux systems programming, kernel development, or another low-level area.
• Strong programming skills in at least one systems language, such as Rust, C, C++, or Go.
• Working knowledge of virtualization or container-isolation technologies such as KVM, QEMU, Firecracker, gVisor, or runc.
• Understanding of Linux resource management, including namespaces, cgroups, scheduling, and memory management.
• Experience profiling and debugging system-level performance issues.
• Ability to work in unfamiliar codebases and technical areas and deliver practical solutions with measurable results.
• Strong communication skills and a commitment to building safe, reliable systems.
• A bachelor's degree or equivalent combination of education, training, and experience in a relevant field.

Preferred Qualifications

• Experience with multi-tenant sandboxes, serverless platforms, or code-execution platforms.
• Experience with microVMs, user-space kernels, lightweight virtualization, snapshot and restore, OS-level security hardening, seccomp, LSMs, or threat modeling of untrusted code.
• Experience with eBPF, kernel tracing tools, Kubernetes, AWS, GCP, network isolation, high-performance networking, egress proxies, or policy layers.
• Experience with sandboxes for AI agents, computer use, or reinforcement-learning environments.
• Experience with GPU virtualization, accelerators, or machine-learning workload performance.
• Prior experience as a technical lead or mentor.

Representative Projects

• Reducing sandbox cold starts from seconds to milliseconds with snapshot-based restore and pre-warmed pools.
• Designing network egress policy layers that allow agents to access approved services without exposing secrets.
• Running thousands of parallel reinforcement-learning environments with strong isolation.
• Building tracing tools that attribute latency and resource use inside running sandboxes.
• Reducing per-sandbox memory overhead to improve host density.
• Hardening runtimes against escape attempts identified through internal red-team exercises.

Benefits

Anthropic offers competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and office spaces for collaboration.