AI Red Teamers: Edit and Label Agent Conversations

We're hiring technical experts to red-team AI models and help build a rigorous safety dataset. This work directly influences how tool-using agents handle complex, multi-turn interactions. Your efforts ensure these models behave safely and predictably in challenging real-world environments.

You will interact with tool-using AI agents through a remote platform to test their safety boundaries. During these sessions, you will drive long conversations that attempt to trigger specific failure modes or edge cases. You will then carefully edit the resulting dialogues and apply structured labels to document the agent's behavior. This process requires deep focus as you guide the model through complex scenarios and evaluate its responses step by step.

We are looking for technical professionals with a strong understanding of AI systems and conversational flows. We welcome software engineers, data scientists, AI researchers, and technical writers with experience in prompt engineering or red-teaming. Candidates should be highly analytical and comfortable navigating complex logic to evaluate model safety.

• Drive multi-turn conversations with tool-using AI models to test their safety constraints.
• Edit model responses and user prompts to create high-quality training dialogues.
• Apply specific labels and categories to document agent behavior and failure modes.
• Navigate complex logical scenarios to deliberately push the boundaries of model capabilities.

• Professional background in software engineering, data science, or technical writing
• Familiarity with AI safety concepts, red-teaming, or advanced prompt engineering
• Ability to craft detailed, logically complex conversational flows
• Comfortable analyzing and categorizing nuanced text responses

• You will be engaged as an independent contractor.
• This is a fully remote opportunity that can be completed on your own schedule.
• Opportunities can be extended, shortened, or concluded early depending on needs and performance.
• Your participation will not involve access to confidential or proprietary information from any employer, client, or institution.
• Payments are processed weekly based on services rendered.
• We are unable to support H1-B or STEM OPT candidates at this time.