logo

Anthropic

Software Engineer, Systems - Claude Code

Department
Engineering
Job Type / Location
San Francisco
Experience Required
5+ years
Posted On

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.

At Anthropic, we believe new AI capabilities are best achieved through secure foundations, not in spite of them. As capabilities grow more advanced, it is critical that progress moves forward safely and for the benefit of all society. It is the reason why security sits at the center of our work, and not as an afterthought.

We're looking for a systems-oriented Software Engineer to push the performance and reliability of Claude Code and the JavaScript runtime that powers it. Claude Code is our terminal-first, agentic coding tool, and a huge part of what makes it fast—and what lets us ship quickly—is Bun. This role sits at that intersection: you'll work deep in the runtime, the kernel interfaces it sits on top of, and the agent loop on top of it, making everything faster, leaner, and more reliable. Your work will also flow through to the Claude Code Agent SDK, Cowork, and other Labs projects that build on the same foundations.

This is low-level work. We're looking for someone who has spent years writing production C, C++ and Rust, who thinks in syscalls and allocators, and who gets excited about shaving microseconds and megabytes off hot paths.

Responsibilities

  • Design, build, and optimize core systems inside Bun and Claude Code—runtime internals, I/O paths, memory management, and the agent execution loop
  • Drive down memory usage, latency, and startup time across Claude Code and the Bun runtime
  • Dig into performance problems across the stack: profiling, tracing, and fixing issues anywhere from the kernel interface up through the JS engine to the agent layer
  • Collaborate with researchers and product engineers to make sure model capabilities translate into a fast, reliable developer experience
  • Contribute to the Bun runtime and toolkit, which underpins Claude Code, the Agent SDK, Cowork, and other internal Labs products
  • Stay ahead of advancements in runtimes, compilers, and AI-assisted tooling—experimenting, measuring, and bringing the best ideas into our stack

You may be a good fit if you

  • Have a strong background in C and/or C++ and Rust (Zig also welcome)
  • Have deep systems programming experience: syscalls, Linux internals, I/O (kqueue, epoll, io_uring), manual memory management, and performance engineering
  • Have worked on language runtimes, virtual machines, compilers, or JS engines like JavaScriptCore or SpiderMonkey
  • Have 5+ years of work experience

View Assessment Process

Think you'll be a good fit?