Engineering Manager, Agent Security Partnerships

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 growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.

About the role

Agent Security's mission is to safely increase how much of the world's work agents can do, starting inside Anthropic.

Claude now writes and ships code, runs experiments and acts on people's behalf, for hours at a time. How much more it can take on depends on how safely it can operate. If, in the worst case, an agent misbehaves, Anthropic and the public may bear the consequences. Our mandate is to mitigate this risk, and Anthropic’s mission depends on it.

The Partnerships team pairs security engineers with the teams at Anthropic that build Claude, build agents and rely on agents. Those teams are our partners. When a team wants its agents to do more, we threat model the plan with them and build the controls together. Most of this has no playbook, so we have often created new controls from first principles. The ones that work become part of our toolbox, spread to other teams, and in some cases ship in our products.

We're hiring an Engineering Manager to lead this team of security engineers. This is first and foremost a people leadership role. Few have secured agents at scale, especially at a frontier lab, so we don't expect you to have done it. We do expect you to have led security engineering teams, and we care most about how you lead people and reason about security.

This role operates at the frontier. Each model generation, and each new harness, brings new challenges. The number of teams building on agents keeps growing, and partner teams won't always agree with your calls. In return, you will have the chance to directly contribute to making transformative AI go well, which is the heart of our mission.

Key responsibilities

• Hire, mentor and grow senior and staff security engineers, and raise the bar as the team scales
• Set the team's investment strategy, deciding which teams get embedded engineers, and for how long, as demand outgrows the team
• Support the team through high-stakes safety and security calls made under uncertainty and on tight timelines
• Balance long-term investment in shared controls against partner teams’ immediate asks
• Deliver large security migrations across all of Anthropic
• Work through difficult tradeoffs with partner teams in ways that build trust and understanding

Representative Projects

In the past, the team has:

• Invented a new kind of classifier-based control for agents partnering with an internal team to develop and roll it out
• Co-built a sandbox with a product team, choosing what to enforce first so that adoption stuck
• Defined the security standards for running agents across the company and partnered with many different teams on how to hit these standards

Minimum qualifications

• Excellent people management skills. You've hired, developed and managed senior and staff engineers, grown engineers into staff or lead roles, and kept a team delivering through changing priorities.
• Deep security management experience. You've managed security engineering teams and answered to executives for risk and tradeoffs.
• Strong written and oral communication skills. You've explained hard tradeoffs to executives and to engineers in plain words.
• Resilience, emotional awareness, low ego and high empathy
• Experience leading large migrations and highly cross-functional work
• A willingness to dive into the details to tie-break complex trade-offs without any clear answer, and the ability to stand by these decisions

Preferred qualifications

• Experience embedding security engineers with product, infrastructure or research teams, or running a security partnerships team
• Agent security or AI/ML security experience
• Familiarity with AI safety concepts and research
• Leading in hypergrowth