Adversarial Prompt Specialist for AI Model Evaluation

Adversarial Prompt experts engage in project-based work aimed at probing large language models for potential failure modes and harmful outputs. This role involves crafting prompts and scenarios designed to test model guardrails, exploring innovative methods to bypass restrictions, and systematically documenting the outcomes of these tests. You will adopt an adversarial mindset to identify weaknesses in the models, collaborating closely with engineers and safety researchers to share your findings and enhance system defenses. Key Responsibilities

• Develop prompts and scenarios to evaluate the robustness of large language models.
• Investigate and document failure modes and harmful outputs.
• Collaborate with engineers and safety researchers to improve model defenses.
• Explore creative strategies to bypass existing restrictions.

• Experience with large language models and their applications.
• Strong analytical skills and an understanding of adversarial testing.
• Ability to document findings clearly and effectively.
• Excellent collaboration skills to work with technical teams.

Part-time remote position with flexible hours.

Up to $65/hr (depending on the project).

Open to applicants with relevant experience in AI and machine learning.