Member of Technical Staff, Frontier AI

• Own research and evaluation initiatives end-to-end: problem framing, data design, quality calibration, and signal validation.
• Design ML-oriented data systems, including task definitions, annotation schemas, rubrics, incentives, and pipelines optimized for downstream model performance.
• Analyze model and system failures to identify root causes, edge cases, and opportunities for improvement.
• Translate ambiguous, real-world behavior into structured evaluation frameworks and new data categories.
• Work closely with researchers and domain experts to calibrate quality early and continuously raise the signal bar.
• Iterate rapidly on evaluations, datasets, and feedback loops to improve system performance.
• Act as a quality gate: block claims, pause work, or force scope changes when signal strength or data integrity is insufficient.
• Partner with cross-functional and client-facing teams to translate research progress into clear, credible narratives grounded in evidence.
• Identify gaps in data or evaluation coverage and recommend where to invest, iterate, or stop based on learnings and impact.

• Strong judgment around research signal quality and when work is (or is not) ready to be externalized.
• Experience designing ML-oriented datasets, evaluation frameworks, and QA processes.
• Ability to translate messy, real-world system behavior into structured research and evaluation opportunities.
• Comfort operating in ambiguity, with a bias toward ownership and decisive action.
• Clear written and verbal communication, especially when explaining tradeoffs, limitations, and signal strength to technical and non-technical stakeholders.
• Proven ability to work directly with experts during project kickoff, calibration, and iteration.
• A systems-level mindset, with interest in improving end-to-end model or agent performance rather than isolated components.

• Experience with reinforcement learning environments, simulators, or feedback-driven training systems.
• Experience improving agentic systems or AI systems operating in real-world workflows.
• Prior work embedded in applied research or production environments with direct impact on deployed systems.
• Experience with evaluation design for complex or real-world tasks.
• Familiarity with expert incentive design and engagement in high-stakes technical projects.

RequiredAlso useful

Project and contracting process: micro1. Applications continue on the provider's website.