Description
We're seeking a Staff Product Manager to join our GenAI Safety team within Trust & Safety. As a key member of our team, you will define and drive the product strategy for ensuring Pinterest's GenAI-powered systems are safe, fair, and trustworthy.
As a Staff Product Manager, you will be responsible for building proactive safety frameworks that scale with our growing AI capabilities, partnering deeply with engineering, policy, data science, and design to protect our users while enabling Pinterest to innovate responsibly.
This is a high-impact role for someone who is passionate about the intersection of AI and user safety, and who thrives in ambiguous, fast-evolving problem spaces. You will work at the frontier of responsible AI - anticipating novel harms before they emerge, red-teaming new AI features, and translating complex policy goals into measurable product requirements.
Key responsibilities will include:
- Defining and driving the product roadmap for GenAI safety across Pinterest's AI-powered surfaces
- Leading proactive identification of risks, failure modes, and adversarial attack vectors across AI systems
- Designing structured red-teaming exercises and evaluation frameworks before and after product launches
- Partnering closely with Trust & Safety policy, legal, and ethics teams to translate nuanced content guidelines into precise, buildable product requirements and model guardrails
- Working with engineering, ML, design, data science, policy, legal, comms, and operations teams to define, align, and ship AI safety solutions across global markets and diverse user populations
- Defining and tracking quantitative safety metrics, including fairness audits, false positive/negative rates, disparate impact analysis, and content harm reduction
- Developing and maintaining AI safety incident runbooks and escalation frameworks, and leading rapid triage and remediation when AI systems produce harmful or unexpected outputs
If you're passionate about the intersection of AI and user safety, and thrive in ambiguous, fast-evolving problem spaces, we encourage you to apply.