Sr. Data Scientist, Responsible AI

Tech Stack

Responsibilities

• Design and develop automated adversarial testing methodologies, including single-turn, multi-turn, and multimodal attack strategies, to identify vulnerabilities in Generative AI products.
• Build and calibrate hybrid evaluation pipelines combining LLM-based judges, classifiers, and rule-based systems for detecting safety violations, policy breaches, bias, and representational harms.
• Develop and operationalize harm taxonomies grounded in industry standards and Pinterest's Responsible AI and Trust & Safety threat models.
• Design adaptive refinement loops that learn from attack outcomes to iteratively surface deeper and previously unknown vulnerabilities.
• Apply scientific rigor and statistical methods to AI safety evaluation, including benchmark dataset construction, evaluation calibration, and success-metric definition.

Benefits

• Equity
• Health Insurance

Culture

Requirements