Agent Post-Training, Artifacts Research

Highlights

• Train frontier agents shipped in Codex, ChatGPT and the API.
• Teach models to turn vague goals into polished artifacts.
• Shape next-gen agents from training runs into shipped products.

Ideal candidate

High-agency researcher-engineer excited by open-ended agent problems who cares about useful, reliable product behavior and moves fast from vague behavioral issues to shipped model improvements.

Requirements

• Strong fundamentals in ML, software engineering, systems, statistics or related field
• Hands-on experience with LLMs, RL, post-training, evals or production ML systems
• Ability to turn vague behavioral problems into concrete experiments and fixes
• Comfortable working across research, product, infrastructure and safety teams
• Care about product impact and useful, reliable, honest agent behavior