Researcher, Agent Post-Training, Personality

📋 Description

• Develop a rigorous understanding of what makes an agent a great collaborator.
• Turn judgments about model behavior into hypotheses, evals, graders, training.
• Study user signals to understand trust, satisfaction, and outcomes.
• Work with experts to produce high-quality rollout data and evaluations.
• Improve reward models and RL objectives for model behaviors.
• Partner with ChatGPT, Codex, and other teams to validate improvements in real workflows.

🎯 Requirements

• Think from the user’s perspective and care about how models feel.
• Translate subjective product questions into falsifiable hypotheses and evaluations.
• Preserve individuality, adaptability, and behavioral diversity.
• Shape how frontier agents communicate and build trust.
• Strong foundations in ML, software, stats, HCI; quick to learn stack.
• Experience with LLMs, post-training, RL/RLHF, reward modeling.