Frontier AI Security Engineer – Cyber & Autonomy

AI Security Institute; United States; Hybrid; USD 87,000 - 195,000; Full time

Benefits offered by this job

Pre-release model access

Operational support

Hybrid working

Pension contributions

Conference funding

Job summary

AI Security Institute is seeking exceptional Software Engineers to work on frontier AI security in London. You’ll build state-of-the-art evaluations, run pre-deployment tests, and validate model behaviors while collaborating with researchers and engineers who care deeply about impact and autonomy.

You will contribute to cyber ranges, sandbox services, and data preparation for publications, with mentorship from technical leads and opportunities to publish externally.

Qualifications

• Deep Python experience, with a broad ecosystem familiarity.
• Ability to design and evolve software systems or architecture.
• Strong interest in improving AI safety and governance.
• Take ownership of ambiguous, technically hard problems and see them through.
• Clear communication and mentoring to raise engineering standards, with cybersecurity or adversarial-security focus.
• Experience building, deploying, and operating AI evaluations in production environments.
• Familiarity with AI misuse and loss-of-control threat models.

Responsibilities

• Develop analysis tooling to turn results into insights.
• Lead testing exercises and communicate findings and recommendations.
• Deliver engineering work from design through maintenance of production components.
• Write production-quality Python software with testing, docs, and observability.

Skills

Python

Software architecture

AI safety

Ownership & follow-through

Clear communication

AI evaluations

DevOps

AI misuse threats

Job description

AI Security Institute is seeking exceptional Software Engineers to work on frontier AI security in London. You’ll build state-of-the-art evaluations, run pre-deployment tests, and validate model behaviors while collaborating with researchers and engineers who care deeply about impact and autonomy.

You will contribute to cyber ranges, sandbox services, and data preparation for publications, with mentorship from technical leads and opportunities to publish externally.