Zillow
Principal Product Manager, Directed Workflows
$169K–$285K
Remote-USA · Remote · Principal · Full-time · Posted Sep 11, 2026
- Experience
- 10+ years of product management experience; or an equivalent combination of education and experience.
- Education
- Bachelor's degree in Computer Science, Engineering, Business, or a related field; or an equivalent combination of education and experience.
About the job
Leads the strategy and delivery of AI-directed workflows for Zillow Home Loans, defining service processes and building agent, evaluation, and experimentation systems across customer engagement and loan production.
What you'll do
- Set the multi-year product vision and define sales and service processes with business partners.
- Design the AI harness and evaluation infrastructure, using operator feedback and experiments to improve agent decisions and outcomes.
- Prove workflow patterns through pilots, scale successful approaches, and align customer, agent, and operator experiences across Zillow.
What they ask for
- 10+ years of product management experience, including complex systems spanning multiple teams and customer or operator journeys.
- Demonstrated experience shipping agentic or decisioning products as a closed loop, including AI harness and evaluation infrastructure.
- Experience shaping sales or service strategy with business leaders and connecting process design to customer value and growth.
- Strong analytical and stakeholder-influence skills, including using data, experimentation, and evaluation to improve outcomes.
Nice to have
- Mortgage or lending experience.
- Familiarity with workflow platforms, operator surfaces, or capacity-aware routing.
Pay
$178,300.00 - $284,700.00 annually (base salary); $169,300.00 - $270,500.00 annually (base salary)
Tools
AI agents, AI harness, context engineering, tool design, memory, orchestration, guardrails, eval infrastructure
AI focus
Designs AI agents and their harness to determine next actions in customer and loan workflows; builds evaluation infrastructure using operator feedback and experiments to assess and improve agent decisions.
Model-written summary · the original posting is authoritative