Model Engineer

Quick Facts

• Role: Model Engineer (Research-driven RL / training research)

Description

Own ambiguous, hard research problems end-to-end: form hypotheses, design experiments, build the training/eval/data needed to test them, and push results into the next model. You will run rigorous experiments to create measurable improvements, improve data recipes, define good RL data and rewards (including grader design and preventing reward hacking), and advance training science for coding and computer-use agents. You will also build evals and targeted tasks to expose weaknesses and partner with engineers to scale promising work to frontier-scale training runs.

Responsibilities

• Design and run rigorous experiments to convert research ideas into measurable model-capability gains

• Improve data recipes via data quality, deduplication, and mixture experiments

• Define what makes RL data and rewards good, including grader design and preventing reward hacking

• Advance training science (architecture/scaling experiments and post-training recipes for coding and computer-use agents)

• Build evals and targeted tasks to reveal model weaknesses and guide future investments

• Partner with engineers to scale small-experiment results to frontier-scale training runs

Requirements

• Production-quality coding experience in Python and one or more of: PyTorch, JAX, Rust, C++

• Ability to handle ambiguous research tasks with little guidance and own direction end-to-end

• Strong reinforcement learning and machine learning fundamentals

• Onsite requirement: able to work in person at the Palo Alto office

• MS or PhD in CS, ML, math, physics, or related field, or equivalent research experience

Benefits

• Equity

• Comprehensive medical, vision, and dental coverage

• 401(k) retirement plan

• Short & long-term disability insurance

• Life insurance

• Various other discounts and perks

• Annual salary range: $160,000 - $250,000 USD