Openclaw Safety
TextRLHF
Engaged in an Outlier project focused on evaluating and preference-ranking AI agent execution traces (OpenClaw), with a strong emphasis on AI Safety. Key responsibilities include conducting comprehensive reviews of reasoning steps and function calls, assessing multiple model responses, and ranking them based on strict safety guidelines. The role requires identifying potential risks (e.g., toxicity, bias, or PII leakage), evaluating response helpfulness, and writing detailed, well-reasoned justifications for each ranking decision. Consistently maintained high-quality standards through rigorous audits to ensure data accuracy for training future Large Language Models (LLMs).
Engaged in an Outlier project focused on evaluating and preference-ranking AI agent execution traces (OpenClaw), with a strong emphasis on AI Safety. Key responsibilities include conducting comprehensive reviews of reasoning steps and function calls, assessing multiple model responses, and ranking them based on strict safety guidelines. The role requires identifying potential risks (e.g., toxicity, bias, or PII leakage), evaluating response helpfulness, and writing detailed, well-reasoned justifications for each ranking decision. Consistently maintained high-quality standards through rigorous audits to ensure data accuracy for training future Large Language Models (LLMs).
2026 - Present