Intern for AI Security RL Gyms
Core
Designing and validating adversarial prompt injection scenarios in YAML to stress-test AI agents for safety policy violations.
Role type
Intern, AI Security Researcher (Adversarial Evaluation)
Builds
Adversarial training environments (RL Gym) for AI safety evaluation
Domain
AI Safety, Adversarial Machine Learning, Cybersecurity
Deliverable
production ML models
Required skills
Prompt injection concepts, YAML, Command Line Interface (CLI), Problem-solving, Attention to detail
Preferred skills
Docker, Computer Science, Cybersecurity, AI, Software Engineering
Technologies
YAML, Docker, CLI
Responsibilities
Design prompt injection scenarios in YAML, Run scenarios against frontier models, Validate success rates of adversarial tasks, Submit passing tasks to the platform
