Background

I spent nearly four years at Box, most recently as Director of Product Management for the AI, agentic, and ML platforms. Before that, I was their Group Product Manager for Search and AI.

• Launched Box AI, the flagship product behind Box's customer-facing AI features.

• Built the AI platform that supports search, machine learning, and AI across Box's teams.

• Launched an Agents platform for building agents on top of Box AI.

• Ran an evaluation program that benchmarked new external models the day providers released them, using proprietary tests built from real enterprise use cases, and gave model providers early feedback.

• Advised C-level and SVP leaders at Box customers on AI and agent use cases.

• Managed a team of product managers.

Before Box, I led data and machine learning products at Zulily and Verizon Media and developed taxonomies for Amazon and Payscale.

I have a Master of Arts in Library and Information Science and a Master of Science in Organic Chemistry, both from the University of Washington.

Why this background matters for your AI initiative

• Content and metadata decide whether an agent works. An AI system can only use what it can find, read, and is allowed to see. Years in taxonomy and search taught me where to look when a pilot stalls on content.

• I write the bar down before I build. A/B testing search relevance at Zulily and benchmarking new models against real enterprise cases at Box are the same discipline: define what "good enough to ship" means, then test against it. A common reason pilots stall is that nobody wrote that bar down.

• I've worked on both sides of the data science handoff. I've directed data science teams, built machine learning pipelines, and shipped the product on top of them. That handoff is where many pilots break.