Research Engineer, Codex

Highlights

• Train the frontier agent models shipped in Codex and ChatGPT.
• Define next-gen agent abilities and run experiments to make them real.
• Carry new capabilities through training runs into shipped products.

What they offer

• Equity

Ideal candidate

High-agency researcher-engineer who thrives on ambiguous problems and cares about shipping useful, reliable agents. Moves fluidly across research, engineering, and product to turn behavior insights into real model improvements.

Requirements

• Strong technical fundamentals in ML, software engineering, systems, statistics or related field
• Hands-on experience with LLMs, RL, post-training, evals, coding or tool-using agents, or production ML
• Excited by open-ended problems with noisy signal requiring research taste and engineering execution
• Care about product impact and what makes an agent useful, reliable and easy to work with
• Can move from vague behavioral problem to concrete experiment and decide next steps
• Comfortable working across research, product, infrastructure, data, evals and safety
• Willing to build load-bearing systems and processes when the team needs it
• Want to train and ship models that make agents genuinely useful