AI Red Team Engineer

We are currently building Watcher, a monitoring tool for coding agents. Our monitoring research agenda attempts to translate compute into safety at scale. Red teaming previously sat inside the RS (Control) role as a partial responsibility. As it’s grown from a single pilot into a recurring need, it now needs a dedicated owner.

As the AI Red Team Engineer you will help build the practice of red teaming AI monitors (both Watcher’s own defenses and frontier labs’ monitoring systems (see our pilot campaign red teaming Anthropic’s auto mode)). You will hunt for attack surfaces monitors that haven’t been tested against yet and turn what you find into fixes.You’ll work closely with Marius (CEO & currently leads the monitoring efforts), control researchers and product engineers.

You will like this opportunity if you think like an attacker and want your adversarial findings to directly strengthen AI monitoring systems. You will join a small team and will have significant ability to shape the team & tech, and have the ability to earn responsibility quickly., * Design and run red teaming campaigns combining severity graded failure mode injection into real trajectories, static monitoring benchmarks (e.g. MonitoringBench), and dynamic off policy control red teaming.

• Identify novel attack surfaces monitors haven’t been tested against.
• Build and maintain automated red teaming pipelines that attack monitors at scale, rather than relying on one off manual probing.
• Design iterative adversarial red team/blue team games, working with RS (Control) on the blue team side to keep escalating attack difficulty as monitors improve.
• Track the research literature and real world incidents on agent failure modes, monitor evasion, and adversarial robustness.
• Turn each campaign into specific, actionable recommendations for monitor developers e.g. on time of intervention, user authorisation, trust boundaries, and monitor visibility
• Write up campaign findings as external publications and internal reports.
• Feed findings back into Watcher’s monitor development and Apollo’s blue teaming research, so lessons learned externally harden our own product., * Run a red teaming campaign against a frontier lab’s monitor : work through real coding sessions with injected attacks across a range of failure modes, identify improvement areas, and deliver actionable recommendations the partner lab can implement.
• Build an automated red teaming pipeline that attacks Watcher’s monitors at scale, rather than relying on one off manual red teaming.
• Investigate a new attack surface : an emerging agent capability or novel evasion vector not yet in Apollo’s failure mode catalog and produce a write up and recommendations for the monitoring team., * Location : This is an in person role working out of our London or San Francisco office. We offer flexible working hours and wfh arrangements.
• Visa sponsorship : We sponsor visas in both the UK and US. Sponsorship isn’t guaranteed for every role or candidate, but if we make you an offer, we’ll work with you to find the right visa route., The product team consists of research scientists: Victor Gillioz, Monika Jotautait , Dmitrii Volkov; product engineers: Jeremy Neiman, Zak Walters, Zen van Riel, Srdjan Miletic and Gustavo Bicalho; and our GTM lead: Kyle Dai. Marius Hobbhahn (CEO) advises the team. Furthermore you will interact with our other SWEs and researchers, since we intend to be our own customer by using our products internally for our research work. You can find our full team here.

Must-haves

• 2+ years of experience in offensive security, adversarial ML, or red teaming of AI systems.
• Strong experience with AI coding agents, extensive use of, comparison of, or direct development of frontier coding agents.
• Experience designing and executing structured adversarial testing campaigns
• Strong Python programming skills
• Strong written communication as you will produce publications and campaign write ups that need to be clear and credible externally.
• Demonstrated ability to work independently on open ended adversarial problems.

Good-to-haves

• Familiarity with AI safety concepts, particularly agent related risks
• Experience with LLM as a judge setups or AI monitoring more broadly
• Background in penetration testing, CTFs, or computer security more broadly.

We want to emphasize that people who feel they don’t fulfill all of these characteristics but think they would be a good fit for the position nonetheless are strongly encouraged to apply. We believe that excellent candidates can come from a variety of backgrounds and are excited to give you opportunities to shine.

• This role offers market competitive salary, equity, and competitive benefits.
• Salary: 100k - 200 GBP (~150k - 270k USD)
• Flexible work hours and schedule
• Unlimited vacation
• Unlimited sick leave
• Up to 6 months of paid parental leave
• Comprehensive health, dental and vision insurance
• Retirement savings with competitive employer matching (e.g. 401(k) for US employees)
• Lunch, dinner, and snacks are provided for all employees on workdays
• Paid work trips, including staff retreats, business trips, and relevant conferences
• A yearly $1,000 (USD) professional development budget
• Relocation support and visa fees (if applicable)

The rapid rise in AI capabilities offers tremendous opportunities, but also presents significant risks. At Apollo Research, we’re primarily concerned with risks from Loss of Control, i.e. risks coming from the model itself rather than e.g. humans misusing the AI. We’re particularly concerned with deceptive alignment / scheming, a phenomenon where a model appears to be aligned but is, in fact, misaligned and capable of evading human oversight.

We work on the science of scheming, detection of scheming (e.g. building evaluations), and scheming mitigations (e.g. anti scheming). We also work on control and monitoring research (see our scalable monitoring agenda). We work closely with many frontier AI companies, such as OpenAI, Anthropic, Google, Meta, Thinking Machines and others, e.g. to test their models and collaborate on the science of scheming. At Apollo, we aim for a culture that emphasizes truth seeking, being goal oriented, giving and receiving constructive feedback, and being friendly and helpful. If you’re interested in more details about what it’s like working at Apollo, you can find more information here.

We also build a coding agent security product called Watcher that secures agent deployments in companies. Our goal is to reduce the probability of catastrophic incidents by securing coding agents, learning about their real world risks, and publishing our research on how to build these control systems most effectively.