Model Behavior Engineer, Reasoning and Decision Making

Role overview

Improve AI systems that help people make difficult decisions when evidence is incomplete and the consequences are significant. You’ll investigate real decision-making needs, evaluate how well agents reason, and turn findings into product improvements alongside engineering, product, design, and subject-matter colleagues. Early work focuses on scientific research and commercial strategy, with methods that can extend to other domains.

Responsibilities

- Study decisions with users and domain experts to clarify goals, constraints, evidence, and unresolved questions. - Create evaluations for complex reasoning tasks, combining close review of examples, expert judgment, and quantitative measures. - Investigate system failures and test changes to how agents structure problems, gather evidence, use tools, retain context, and revise conclusions. - Design software components and data structures for reasoning workflows and evaluations, balancing reliability, flexibility, speed, and cost. - Ship and maintain improvements, participate in technical reviews, and monitor how systems perform in real use.

Requirements

- Strong analytical judgment: make ambiguous questions precise, identify assumptions, weigh conflicting evidence, and explain what could change your view. - Ability to make progress in unfamiliar subject areas and communicate clearly with people who have different expertise. - Experience designing useful experiments when there is no established benchmark, including examining failures and spotting misleading measures. - Ability to write and debug code independently, work with data and model APIs, and carry experiments into maintained software; Python or a comparable language is suitable. - Practical familiarity with language models and agent workflows, plus sound software design and engineering judgment.

Benefits and work setup

- Work remotely or from an office in Oakland, California; team retreats and coworking events involve occasional in-person travel. - Benefits include health, dental, vision, and life insurance, flexible vacation with a 20-day annual recommendation, a retirement plan with employer match, and budgets for wellbeing, equipment, and AI learning. - Target starting compensation varies by level: $185,000–$220,000 for career-level hires, $220,000–$260,000 for senior hires, and $250,000–$320,000 for staff-level hires, plus equity.