Staff+ Software Engineer, Safeguards ML Infrastructure

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

The Safeguards ML Infra team designs, builds, and operates the production infrastructure that powers Claude's safety systems. We own both the critical backend services that ensure safety on the token generation path, as well as the infrastructure that configures these systems during model provisioning for every platform Claude runs on -- 1P, Bedrock, Vertex, and beyond. We define and maintain SLOs, build the observability systems that surface problems early, and lead incident response when issues arise.

We're growing the team and looking for Software Engineers with deep experience owning production infrastructure at scale. The ideal candidate has built and operated large-scale distributed systems under real production pressure and built platforms, tooling, and infrastructure that other engineers depend on. Familiarity with ML research or transformer architectures is not required -- you will learn that on the job.

What we prioritize is distributed backend systems expertise and a track record of ownership over the production environment.

Responsibilities

: Design, build, and deploy backend services that are critical safety pieces on the token sampling and generation path. Own and operate the production serving infrastructure for those services across multiple deployment platforms (1P, AWS Bedrock, GCP Vertex). Define and maintain SLOs, build observability and alerting systems, and lead incident response for infrastructure on the critical path of every Claude request Participate in on-call and operational-duty rotations covering service incidents, model provisioning, and time-sensitive research and safety launches.

Reduce oncall and onduty toil by building automation, tooling, and self-serve workflows that minimize manual operations. Be the first user of the systems you build, running them for real workloads yourself before other teams depend on them. Build and maintain a safety registry with full provenance -- tracking what is running in production, on which model, and when and by whom it was deployed.

Implement automated post-deploy validation to ensure correctness is consistent across platforms. Work closely with ML researchers to productionize new safety techniques, translating experimental work into reliable, scalable production systems. Contribute to the long-term goal of platform-agnostic deployment tooling that brings 3P platforms to parity with 1P operational maturity.

You may be a good fit if you: Are proficient in Python; experience with Rust is a plus but not required. Have designed, built, and operated high QPS systems at global scale. Have a strong foundation in distributed systems: replication, consistency tradeoffs, failure modes, and SLO management under load.

Have meaningful on-call experience for production systems, including incident response and postmortem-driven improvements. Have a desire to close the gap where nobody has yet raised their hand, even if it requires manually hand holding processes until automation and tooling can be built. Have hands-on experience deploying and operating on cloud platforms (AWS, GCP) at scale Approach infrastructure as a platform -- building systems and abstractions that other engineers build on, rather than point solutions for a single team's needs.

Strong candidates may have experience with: Have 8+ years of industry software engineering experience Experience building deployment and rollout systems with canary analysis, automated validation, or progressive rollout controls.