jobs

Agent Post-Training Research

OpenAI’s Agent Post-Training team is developing the frontier agents behind Codex, ChatGPT, the API, and other products. These systems are designed to operate computers, use tools, collaborate with people and other agents, and complete long-

Company

OpenAI

Location

San Francisco, United States; London, United Kingdom

Status

Closed

Posted

2026-09-13T13:31:38.557+00:00

OpenAI's Agent Post-Training team works on the agentic models behind Codex, ChatGPT and the API, teaching systems to operate computers, use tools and finish long-horizon tasks. The person improves these models' capability, reliability and product fit — owning a research direction, building large-scale training infrastructure, creating evaluations, or carrying a capability from research into launch. Work spans reinforcement learning, synthetic data, reward signals, graders, tool use and multi-agent coordination. The posting asks for strong machine-learning, software, systems or statistics fundamentals plus hands-on LLM, post-training, evaluation, coding-agent or production ML experience.