Data Scientist - AI Safety

Skills

About the Role

You will own safety datasets from collection through cleaning labeling quality control versioning and training or evaluation readiness. You will translate safety policies into consistent criteria design contributor workflows and build reproducible Python and SQL pipelines. You will also evaluate production models and work across policy engineering and research disciplines.

Requirements

• Experience as a Data Scientist or similar working with real-world datasets and ML models
• Knowledge of dataset collection cleaning sampling labeling quality control and evaluation
• Proficiency with Python and SQL
• Strong critical thinking and judgment
• Ability to work autonomously and communicate across disciplines
• Experience in AI safety trust and safety content moderation or fraud detection is beneficial
• Experience building datasets or evaluation frameworks is beneficial
• Experience managing human-in-the-loop labeling processes or dataset versioning is beneficial
• Experience operationalizing guidelines with policy or legal teams is beneficial

Responsibilities

• Own safety datasets end to end
• Translate safety policy into labeling and evaluation criteria
• Design and manage external contributor labeling processes
• Build production model evaluation workflows
• Develop Python and SQL data pipelines and tooling

Benefits

• Annual discretionary professional development stipend
• Annual discretionary social travel stipend
• Annual company offsite
• Monthly coworking stipend