Senior Staff, Product Manager- AI Evaluation & Quality

ServiceNow is a company focused on AI-driven business reinvention, offering a unique opportunity for a Senior Staff Product Manager in AI Evaluation & Quality. The role involves building an evaluation methodology and continuous improvement platform for AI specialists, ensuring high-quality performance across business units.

Responsibilities:

• Define the end-to-end evaluation methodology across ServiceNow's AI specialist portfolio — golden datasets, LLM-as-judge calibration, failure taxonomy, and a multi-layer metric model spanning agent behavior, user experience, and business impact
• Move evaluation from a one-time release gate to a continuous improvement engine, where production telemetry feeds failure analysis, targeted evaluation expansion, and redeployment — making live signal, not synthetic testing, the primary driver of quality
• Partner with a federated Center of Excellence model — central methodology and tooling, embedded practice in each business unit — acting as the internal consulting function that helps teams stand up evaluation without rebuilding infrastructure from scratch
• Drive evaluation from bespoke effort to reusable, self-serve platform capability — evaluation infrastructure, data tooling, and calibration systems that make closed-loop evaluation available to every team building AI specialists

Requirements:

• 12+ years of software product management experience
• Deep experience in AI/ML product management, ideally with hands-on exposure to LLM evaluation, agentic systems, or applied ML quality frameworks
• A track record of defining methodology or standards in ambiguous, cross-team problem spaces — not just executing an existing roadmap
• Strong technical fluency — comfortable in the details of evaluation pipelines, data platforms, and production telemetry, not just the product narrative around them
• Proven ability to drive outcomes through influence and technical credibility, without formal authority over the teams you're working with