Security Researcher (Research Coding Tasks)

You will create self-contained coding tasks used to test AI models, using your expertise in security research. You will also review model outputs and give feedback based on research-quality standards.

Your work may cover fuzzing, vulnerability detection, binary analysis, systems security, and LLM agent security. You will work with a project lead to keep tasks accurate, rigorous, and connected to current security research.

• Build technical research coding tasks in your security specialty.
• Review AI-generated work and provide clear, technically grounded feedback.
• Develop tasks that reflect your existing research practice.
• Help maintain accurate and reproducible research standards.

This is an asynchronous remote contract designed to fit around your research schedule. The role is open to candidates in the eligible countries for this project.

• Pay: $50 per hour.
• Time: 20 or more hours per week.
• Language: Fluent in English.
• Education: Completed or in-progress PhD in computer science or a related field.
• Research: Published security research with code released through GitHub or a comparable research artifact.
• Expertise: Ability to contribute at research quality in a security specialty.
• Communication: Strong written technical communication and sound judgment when evaluating AI-generated work.
• Helpful background: First-author publications at USENIX Security, CCS, IEEE S&P, NDSS, or comparable venues.
• Helpful background: Depth in fuzzing, vulnerability detection, binary analysis, systems security, or LLM agent security.
• Helpful background: Comfort with Docker and reproducible research environments.

Apply on OpenTrain with your resume, then complete the application on the hiring site.

AI training is the human work behind systems that learn from examples. People create coding tasks, review model answers, and provide feedback so AI systems become more accurate and useful; specialized researchers are paid for applying their field expertise.