Agent Post-Training, Artifacts Research

Responsibilities

• Design and run experiments to improve agentic model behavior for complex software and plugins
• Own end-to-end improvements to the post-training stack, including RL, data pipelines, graders, reward signals, and evals
• Build environments that expose model failures and convert them into training data or new research directions
• Partner with Codex and ChatGPT product teams to translate user needs into model improvements
• Work on early-training and alignment interventions, including data mixtures, synthetic data, and objective functions
• Improve large-scale training machinery, focusing on experiment velocity, reliability, and production readiness
• Debug hard failures in shipped models and translate qualitative behavior into concrete hypotheses and fixes

Requirements

• Strong technical fundamentals in machine learning, software engineering, systems, or statistics
• Hands-on experience with LLMs, RL, RLHF/RLAIF, or post-training techniques
• Experience with synthetic data generation, model training, or building evaluation frameworks
• Proven ability to move from vague behavioral problems to concrete, executable experiments
• Experience working with coding agents, tool-using agents, or production ML systems

Preferred Qualifications

• Background in consulting, finance, marketing, operations, or data science
• Experience working across research, product, and infrastructure boundaries

Benefits

• Competitive base salary and generous equity
• Medical, dental, and vision insurance for you and your family
• 401(k) retirement plan with employer match
• Paid parental leave and flexible PTO
• Daily meals in the office and meal delivery credits
• Annual learning and development stipend
• Mental health and wellness support

About the Company

OpenAI is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity. We push the boundaries of AI capabilities and seek to safely deploy them to the world through products like ChatGPT and the API.