Engineering Manager, Guardian

Description

The Guardian team builds safety and oversight systems for AI agents that are becoming more capable. The aim is to let agents handle longer tasks with more autonomy while keeping real safeguards, following what the user intends, and stepping in when an action could cause serious harm or harm that is hard to undo.

Auto-review is the first version of this work. When an agent hits a permission boundary, a separate model looks at the proposed action in context and decides whether it can go ahead. The team is now building richer oversight that takes more user context into account, supports a wider range of agent experiences, and connects across the company products.

Guardian works across engineering, alignment research, product, infrastructure, and security. The existing engineering group in London is strong and will keep growing as a small, closely working team.

Responsibilities

• Lead, hire, and grow a team of engineers in London working on agent safety, backend systems, applied AI, and product infrastructure.
• Set the technical direction for Auto-review and the wider Guardian platform, weighing safety, product usefulness, latency, and reliability.
• Build production systems that assess agent actions in real time, flag potentially harmful behavior, and enable safe recovery or escalation.
• Improve how oversight systems read user intent, environment, permissions, and relevant personalization signals, while protecting privacy and limiting unnecessary friction.
• Work with alignment researchers to turn prototypes, evaluations, and new monitoring methods into products that scale.
• Work with Product, Security, Infrastructure, and other engineering teams to bring Guardian capabilities into agent experiences across the company.
• Set up strict evaluation and operations practices, including adversarial testing, failure analysis, monitoring, and metrics for safety, false positives, latency, and user impact.

Requirements

• Have led engineering teams that delivered complex, high-stakes production systems from start to finish.
• Have built with AI agents or large language models in a core product, a startup, or a serious independent project, and understand the particular challenges of agentic systems.
• Show strong product-engineering judgment and can balance safety, usability, speed, and reliability.
• Can work well with research teams, turn open-ended findings into engineering plans, and explain ideas clearly across technical fields.
• Have a clear way of managing hiring, coaching, performance, and delivery.
• Care about building AI systems that are useful, trustworthy, and safe.

Conditions

This role is based in the London office.

The company is an AI research and deployment company focused on making sure general-purpose AI benefits all of humanity. We build frontier systems and deploy them through our products.

The company is an equal opportunity employer and does not discriminate on the basis of any legally protected characteristic.