Anthropic

Overview

We are looking for software engineers to help build safety and oversight mechanisms for our AI systems. As a software engineer on the Safeguards team, you will work to monitor models, prevent misuse, and ensure user well-being. This role will focus on building systems to detect unwanted model behaviors and prevent disallowed use of models.

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.

Requirements & Eligibility

• Bachelor’s degree in Computer Science, Software Engineering or comparable experience
• Proficiency in Python, Distributed Systems, and Infrastructure as Code
• Have worked across multiple cloud providers, or built infrastructure designed to be provider-agnostic
• Have built and operated large-scale, distributed infrastructure, such as data platforms, control planes, or job schedulers
• Have experience with sandboxing and isolation technologies (containers, microVMs, network policy) or with a systems language such as Rust
• Strong communication skills and ability to explain complex technical concepts to non-technical stakeholders
• 8+ years of experience in a software engineering position

Key Responsibilities

• Develop monitoring systems to detect unwanted behaviors from our API partners and potentially take automated enforcement actions; surface these in internal dashboards to analysts for manual review
• Stand up and run deployments of those monitoring systems across multiple clouds, including inside cloud-provider partner environments where data must stay in place. Keep the deployments consistent through shared deployment pipelines, smoke tests, obs
• Design and harden the sandboxed runtime that AI agents execute in: isolation, network egress controls, least-privilege data access, audit logging, and insider-risk controls
• Manage cost and capacity for large volumes of long-running agent work, and set and meet service-level objectives for the platform
• Partner with the engineers and researchers who write and evaluate the monitoring agents, and with security, privacy, and legal teams, so new detection work can ship quickly on a platform everyone trusts