Data Scientist, Preparedness

The role

The Data Scientist will build, evaluate, and improve mitigations to prevent extreme harms from AI systems. This role involves creating mitigation intelligence and monitoring systems to detect issues and reduce over-blocking and under-blocking.

What you will do

• Evaluate and improve mitigation systems including classifiers and detection pipelines.
• Diagnose false positives and false negatives with deep error analysis.
• Build monitoring and measurement frameworks to track mitigation effectiveness.
• Identify trends in over-blocking vs. under-blocking and propose interventions.
• Develop insights from customer feedback and usage patterns to detect shifts in adversarial behavior.
• Expand risk monitoring into new areas like cybersecurity threats and model loss-of-control.
• Communicate results to technical and executive stakeholders.

Required qualifications

• Significant experience in data science or applied analytics in high-stakes domains.
• Strong foundations in experimentation, causal thinking, and/or observational inference.
• Fluency in SQL and Python (or equivalent).
• Experience building metrics, dashboards, and operational monitoring.
• Track record of driving cross-functional impact with engineering, product, and research partners.
• Experience with classifier evaluation, calibration, thresholding, and error analysis at scale.
• Genuine interest in AI safety, alignment, and catastrophic risk prevention.

Nice to have

• Cybersecurity data science experience, including exposure to threat modeling, adversarial dynamics, abuse patterns, or security telemetry.
• Trust & Safety experience is helpful, but not required.

About the company

OpenAI is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity. They seek to safely deploy AI to the world through their products.