Staff+ Software Engineer, Inference Velocity

Job summary

Responsibilities

Anthropic's Inference organization serves Claude to millions of users and enterprise customers with the speed, reliability, and efficiency that frontier AI demands. We build across GPUs, TPUs, and Trainium, and the complexity of our development environment grows with every platform we add.

• Set technical direction for Inference Developer Productivity, owning the architecture and roadmap for toolchains, dev environments, and CI/CD across GPU (CUDA), TPU, and Trainium…
• Be the technical owner of accelerator toolchain management: compilers, drivers, libraries, frameworks, kept current, compatible, and well-tested so Inference engineers focus on…
• Design and build infrastructure for efficient accelerator usage during development, including devbox environments, pre- and post-land validation automation, and shared tooling…
• Define and instrument productivity metrics for the Inference org, building the dashboards and alerting that surface regressions early (smoke tests red for extended periods, build…
• Proactively hunt down bottlenecks, toil, and friction across Inference engineering workflows, then design and build the systems that eliminate them

What they're looking for

• 8+ years of software engineering experience, with significant time as the technical lead or anchor on an infrastructure, platform, or developer productivity team
• Deep background in systems engineering, build/test infrastructure, or ML infrastructure, with the ability to go hands-on with toolchain issues, CI/CD pipelines, and developer…
• Experience owning toolchains or development environments for compute-intensive workloads (ML training or inference, HPC, large-scale distributed systems)
• Real depth in at least one accelerator ecosystem (CUDA/GPU, TPU, or Trainium/AWS Neuron) and genuine appetite to learn the others
• A track record of defining and using engineering metrics to drive improvement: you've built dashboards, set SLOs on developer workflows, or led initiatives that measurably…

Tech stack

• CUDA
• AWS
• Triton
• Kubernetes