Agent Post-Training Research

Team Description OpenAI is looking for exceptional researchers to join the Post-Training Frontiers team, which is responsible for post-training the agentic models we ship across Codex, the API, ChatGPT Thinking, and ChatGPT Pro. The Post-Training Frontiers team sets up the pipeline for deciding which integrations can go into the post-training run, develops its own horizontal improvements to the model, and trains the final model. The role requires working on the most impactful horizontal improvements for the next model, which could include factuality, instruction following, function calling, multi-agent collaboration, calibrated reasoning effort, tool use, or improving taste in our models. You might build or improve our grading stack, improve our user-data flywheel, or automate our processes to make large post-training runs faster, more reliable, and easier for researchers to use. This is a team for people who want their work to land directly in models used by hundreds of millions of people. The right person is deeply technical, highly independent, goal-oriented rather than method-oriented, and excited by the messy, high-agency work of turning research ideas into production model behavior In this role, you will: Own end-to-end research and engineering projects that improve the final post-training of OpenAI’s agentic models. Decide, together with partner teams, which integrations are ready for inclusion in major model runs. Develop horizontal model improvements across factuality, instruction following, tool/function calling, multi-agent behavior, reasoning-effort calibration, and other broad capabilities. Build and improve training, evaluation, grading, and data infrastructure for large-scale RL/post-training runs. Create evals and diagnostics that help us understand whether a model is ready to ship. Improve the feedback loop from real product usage into post-training, including better ways to learn from implicit user feedback. Collaborate closely with Codex, API, ChatGPT, product, training, and other post-training teams to make frontier models more useful, reliable, and agentic. You might thrive in this role if you: Have strong ML fundamentals and hands-on experience with LLMs, RL, RLHF, post-training, evals, or model training. Are an unusually strong engineer who can move quickly in complex systems and make pragmatic technical decisions. Can own ambiguous problems end-to-end without needing a tightly specified roadmap. Care more about impact than method, and are happy to do unglamorous but load-bearing work when it matters. Have excellent taste in model behavior and can reason about what “good” looks like across many user-facing domains. Are comfortable working across research, infrastructure, data, evals, and product boundaries. Are excited to train and ship the frontier agentic models that power Codex, ChatGPT, and the API. Nice to have: Experience with large-scale model training or RL systems. Experience building evals, graders, reward models, or data pipelines for LLM training. Experience with coding agents, tool-using agents, browser/computer-use agents, function calling, or multi-agent systems. Background in quant, systems, infra, or other environments where you built reliable machinery for high-stakes experimentation. Evidence of strong product taste, especially around writing, design, code generation, or agent workflows. About OpenAI 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 the capabilities of AI systems and seek to safely deploy them to the world through our products. AI is an extremely powerful tool that must be created with safety and human needs at its core, and to achieve our mission, we must encompass and value the many different perspectives, voices, and experiences that form the full spectrum of humanity. We are an equal opportunity employer, and we do not discriminate on the basis of race, religion, color, national origin, sex, sexual orientation, age, veteran status, disability, genetic information, or other applicable legally protected characteristic. For additional information, please see OpenAI’s Affirmative Action and Equal Employment Opportunity Policy Statement . Background checks for applicants will be administered in accordance with applicable law, and qualified applicants with arrest or conviction records will be considered for employment consistent with those laws, including the San Francisco Fair Chance Ordinance, the Los Angeles County Fair Chance Ordinance for Employers, and the California Fair Chance Act, for US-based candidates. For unincorporated Los Angeles County workers: we reasonably believe that criminal history may have a direct, adverse and negative relationship with the following job duties, potentially resulting in the withdrawal of a conditional offer of employment: protect computer hardware entrusted to you from theft, loss or damage; retu