Research Scientist

What you'll do

• Research data collection strategies and design data slices to reveal model failure modes.
• Model annotator behavior and run experiments to improve instruction clarity and reward signal reliability.
• Develop metrics and frameworks to evaluate dataset quality, diversity, and downstream alignment impact.

What they ask for

• Curiosity about how data structure, quality, and selection influence LLM performance.
• Ability to design lightweight experiments and iterate quickly toward actionable insights.
• Drive to build data foundations that shape future AI systems.

Tools

LLM

AI focus

Designing datasets and data collection strategies for frontier models; evaluating dataset quality and their effects on alignment, safety, and model performance.