AI Safety & Evaluation Tools Engineer

Faangapply; San Francisco (CA); On-site; USD 140,000 - 230,000; Full time

Job summary

OpenAI is seeking a Software Engineer to build the tools and infrastructure that support cybersecurity research, frontier‑risk evaluations, and red teaming. You’ll partner with researchers and engineers to translate evolving questions into practical software—from early prototypes to dependable systems.

Projects may include backend services, evaluation infrastructure, data pipelines, and internal applications for tests, model behavior reviews, and results inspection.

Qualifications

• Strong software engineering fundamentals and production‑grade software delivery.
• Experience in backend development, infrastructure, or data systems.
• Ability to translate open-ended questions into usable software solutions.

Responsibilities

• Design, build, and maintain software that supports cybersecurity research, frontier‑risk evaluations, and red teaming.
• Develop tools and workflows that simplify running evaluations, inspecting results, and investigating failures.
• Build services, APIs, data pipelines, and user interfaces to help researchers work efficiently.
• Turn experimental workflows into reusable, well‑tested, and documented systems.
• Own projects from problem understanding to deployment and ongoing operation.
• Improve reliability, performance, observability, security, and usability as needs evolve.
• Collaborate with researchers, security specialists, and engineers to identify useful problems and make practical technical decisions.

Skills

Software engineering fundamentals

Backend development

Infrastructure

Data systems

Security awareness

Cross-disciplinary collaboration

Job description

OpenAI is seeking a Software Engineer to build the tools and infrastructure that support cybersecurity research, frontier‑risk evaluations, and red teaming. You’ll partner with researchers and engineers to translate evolving questions into practical software—from early prototypes to dependable systems.

Projects may include backend services, evaluation infrastructure, data pipelines, and internal applications for tests, model behavior reviews, and results inspection.