Software Engineer, ML Networking

About the role

Anthropic is looking for a systems‑level Software Engineer who will design and maintain the software that connects its AI accelerators to high‑speed networks. The role focuses on low‑latency, high‑throughput networking stack development and optimization across kernel and user space.

Key responsibilities

• Develop and maintain software that interfaces AI accelerators with the data‑center network.
• Design, implement, and benchmark new network protocols, collectives and RPC mechanisms.
• Debug and optimise kernel‑level networking paths (TCP/IP, XDP, eBPF, io_uring, epoll) and user‑space stacks (DPDK, RDMA, kernel bypass).
• Collaborate with hardware teams to integrate PCIe drivers and SmartNIC functionality.
• Profile and improve performance of distributed workloads, including congestion control and latency reduction.

Required profile

• 5+ years of experience in systems or network programming.
• Bachelor’s degree in Computer Science, Electrical Engineering or equivalent experience.
• Strong debugging mindset for complex, multi‑layered issues.
• Experience with high‑performance computing, telecommunications or embedded systems is a plus.

Required skills

• Deep knowledge of network protocols and stack internals (TCP/IP, XDP, eBPF, io_uring, epoll).
• Proficiency with user‑space networking frameworks (DPDK, RDMA, kernel bypass techniques).
• Systems programming expertise, including lock‑free data structures and NUMA‑aware coding.
• Comfort with or willingness to learn Rust.
• Understanding of ML accelerators, PCIe, and SmartNIC development.

What we offer

• Competitive annual compensation ranging from $280 k to $850 k USD.
• Opportunity to work on cutting‑edge AI infrastructure alongside leading researchers.
• Flexible work environment in San Francisco.