Principal Product Manager, Agentic Evals

Expedia Expedia · Hospitality · Seattle, WA +1

The Principal Product Manager, Agentic Evals will lead the strategy and build-out of Expedia Group’s AI evaluation framework. This role involves defining principles, building tools and workflows for measurement, partnering with engineering and data science on metric design and model evaluation, establishing standards for accuracy, safety, transparency, and performance, and influencing strategic decisions on AI investments and risk management. The goal is to ensure AI experiences meet a high standard of quality, safety, and trust across the company.

What you'd actually do

  1. Define Expedia Group’s AI evaluation framework, principles, and adoption strategy
  2. Build tools and workflows that support reliable, scalable measurement across brands
  3. Partner with engineering and data science on metric design, instrumentation, and model evaluation
  4. Establish clear, durable standards for accuracy, safety, transparency, and performance
  5. Publish evaluation insights, scorecards, and recommendations that shape roadmaps

Skills

Required

  • Bachelor's degree in a technical, quantitative, or related field, or equivalent practical experience
  • 10+ years of product management experience, including significant work with AI, ML, data platforms, or evaluation systems
  • Strong technical depth and experience partnering with DS/ML teams
  • Proven ability to build frameworks, metrics, or tools that scale across large organizations
  • Experience with model evaluation, instrumentation, experimentation, or system reliability
  • Strong written and verbal communication that brings clarity to complex technical spaces
  • Track record of influencing cross-functional leaders and driving alignment

Nice to have

  • Experience building evaluation platforms, ML tooling, or governance frameworks
  • Familiarity with LLM behavior, model evaluation methods, or prompt testing
  • Understanding of trustworthy AI principles, safety evaluation, and failure-mode analysis

What the JD emphasized

  • AI evaluation framework
  • model evaluation
  • trustworthy AI principles
  • safety evaluation

Other signals

  • AI evaluation framework
  • metrics, tooling, and processes
  • accuracy, safety, latency, helpfulness
  • trustworthy AI principles