Agent Post-Training, Computer Use Research

Highlights

• Teach frontier models to operate browsers and desktops reliably.
• Design experiments that improve complex computer-use behavior.
• Ship computer-use capabilities into Codex, ChatGPT and API agents.

Ideal candidate

High-agency builder with strong ML fundamentals who loves open-ended agent problems and cares about shipping useful, reliable, honest models into real products.

Requirements

• Strong technical fundamentals in machine learning, software engineering, systems, statistics or related field
• Hands-on experience with LLMs, RL, post-training, evals, graders, synthetic data or production ML systems
• Excited by open-ended problems with noisy signal requiring research taste and engineering execution
• Care about product impact and model behavior, not just benchmark movement
• Able to turn vague behavioral problems into concrete experiments and decide next steps
• Comfortable working across research, product, infrastructure, data, evals and safety teams