WHAT I DID

• Engineered AI Language Governance: Spearheaded the development of a multi-tier governance system for AI-generated size naming (Size 0–Size 4) across high-traffic merchandising surfaces like Trade Up and Trade Across.

• Reframed Metadata as Conversational UX: Transformed static size labels into a dynamic UX system, treating naming as a conversational problem rather than a metadata task to ensure LLM outputs remained intuitive and human-centered.

• Deciphered Customer Nuance: Developed distinct behavioral definitions for seemingly synonymous terms—such as "Travel Size" (portability) vs. "Compact" (space efficiency) vs. "Individual" (portioning)—to align AI outputs with specific shopping motivations.

• Built Scalable Decision Logic: Introduced decision trees and fallback workflows that enabled LLMs to handle complex categorization, ensuring accuracy across hundreds of product types while reducing manual intervention.

• Solved for Global Scalability: Established linguistic guardrails and translation-ready terminology (e.g., transitioning "Single" to "Individual") to prevent cross-market confusion and maintain consistency across international sizing tiers.

• Operationalized Evaluation Frameworks: Designed a repeatable four-step process—covering default selection, packaging assessment, intent confirmation, and uncertainty fallback—to standardize how AI systems select and deploy sizing language.