Staff+ Software Engineer, Safeguards Data

About the role

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

Claude serves millions of individual users alongside enterprise clients. Safeguards Engineering develops the foundational infrastructure that ensures its safe use: robust access controls, active abuse detection, real-time probes and classifiers within the inference pipeline, offline monitoring for review, and the underlying systems powering it all. Operating at scale and velocity, we continuously adapt to technology whose capabilities and risks evolve weekly.

The Safeguards Data team owns the data lifecycle for Safeguards—routing user data and safety metadata from online systems into offline storage and enabling access for downstream teams. Our goal is to provide privacy-safe access to sensitive data for approved uses while ensuring performance, reliability, and minimal blast radius. We prioritize security by design, embedding controls directly into systems operating across AWS, GCP, Azure, and customer-managed environments.

This is a systems engineering position within a small, newly established team. Rather than managing an existing backlog, you will have the autonomy to own a problem domain, establish its technical vision, and shape the team's trajectory.

Key responsibilities

- Design, build, and operate the stores, pipelines, schemas, and APIs that Safeguards' detection, evaluation, investigation, and review systems read from.

- Build privacy-safe interfaces that allow ML training, evaluation, and investigation workflows to utilize sensitive data without requiring broad access to raw records, ensuring access is tied to purpose and built-in lineage.

- Scale the distributed storage, ingestion, and processing systems operated by Safeguards, maintaining cost efficiency, reliability, and portability across deployment environments.

- Collaborate with investigators, researchers, legal, privacy, and security teams to translate operational requirements into functional mechanisms and systems for safe data utilization.

- Uphold a high operational bar by participating in the on-call rotation, leading incident reviews, and continuously reducing manual operational overhead.

- Lead through technical design by authoring foundational proposals and conducting cross-team architectural reviews.

Minimum qualifications

- Have strong systems fundamentals (processes, threads, RPC, storage, and how systems fail) and have built and operated large-scale, fault-tolerant distributed systems

- Have built and operated systems that handle sensitive or high-value data, with access, privacy, retention and security controls designed in from the start