Selected work

Challenged already-public feature names with localization data and customer research, winning director-level approval to rename them before launch.

Wrote system prompts that grounded ad recommendations in each advertiser's own account data, turning generic guidance into a clear, personalized case for action.

Closed the gap between content principles and measurable quality, building the accuracy judge that verified the assistant met its standards at scale.

Spotted a company-wide gap in how LLM judges got built, then wrote the playbook — and resolved the ownership questions — that became the standard other teams followed.

Won leadership approval for Meta's generative AI creative vision, then led content design for the first three tools that shipped: text generation, image expansion, and background generation.

Caught a missed error-state edge case before implementation, avoiding a costly rebuild.

Facilitated cross-functional workshops to resolve ownership confusion after a reorg and unlock roadmap planning for two teams.

Designed the prioritization and planning model that let a new Growth team run roughly 4x the required experiment volume.

Led strategy workshops to give a newly formed team its first customer-centered mission, vision, and Jobs-to-be-Done framework.