Sr. Data Scientist, Responsible AI

📋 Description

• Design automated adversarial testing for GenAI products.
• Build hybrid evaluation pipelines with LLM judges, classifiers, and rules.
• Develop harm taxonomies aligned with Pinterest's safety threat models.
• Create adaptive loops that learn from attack outcomes to surface vulnerabilities.
• Apply statistical methods to evaluate AI safety, metrics, and coverage.
• Collaborate with ML engineers, Safety, policy, product, and legal teams.

🎯 Requirements

• 5+ years of data analysis in fast-paced, data-driven environments.
• Hands-on AI safety/adversarial ML, red teaming, or trust & safety.
• Deep familiarity with LLMs and generative AI failure modes (bias, prompts, safety).
• Experience building AI eval frameworks (LLM-as-judge, classifiers, benchmarks).
• Strong Python, SQL/Spark, ML pipelines, and large-scale experimentation.
• Familiarity with AI safety taxonomies (OWASP LLM Top 10, MITRE ATLAS) preferred.

🎁 Benefits

• Equal opportunity employer with an inclusive culture.
• Flexible work arrangements (Pinflex) to help you do your best work.
• Pinterest is committed to inclusion and equal opportunity for all.
• Relocation assistance is not offered for this position.