At Handshake AI, I annotated and evaluated more than 100,000 multimodal data points and collaborated with data scientists and machine learning engineers to fine-tune five models. Across evaluation rounds, the models reached 97% precision and recall.
I authored annotation guidelines and edge-case manuals, helping increase annotation speed by 20% while reducing misclassification. I also used Snorkel for weak-supervision labeling, cutting manual labeling cycle times by 30%.
On the Applied AI Research & Visual Grounding Initiative, I developed a four-tier framework for classifying visual-grounding errors in vision-language models. I evaluated edge cases and turned qualitative patterns into quantitative metrics and research briefs.
My work also spans creative production and public-sector communications. I designed WordPress and HTML websites at Rex Graphics, wrote more than 200 pieces as an independent writer, and translated technical material into plain-language policy documents.

