// about
At Nordstrom, I work on multimodal models, agents, and recommendation systems. Most of my favorite problems are about whether something works. For the style filters customers shop with, I set the thresholds. For NordAssist, our agentic assistant, I owned the architecture decision. I also study when a model's judgment can stand in for a person's.
Part of this comes from my work in healthcare. At Weill Cornell Medicine, I built an NLP dashboard for radiologists. At HealthTap, my models flagged risky answers from a network of 90,000+ doctors. In both places, a model's mistake could land on a patient. That's where I learned to ask for evidence before trusting a model. It still shapes how I work.
I have an M.S. from Cornell, focused on AI. I review for ML and NLP workshops. I also moderate generative AI sessions at the AI4 conference.
// selected papers
Sunburst: Open Source Radiologist Timeline Dashboard for Summarizing Prior Imaging Reports Using NLP and UMLS
Research conducted at Weill Cornell Medicine
// news
- Aug '26 Moderated RAG systems session at AI4 2026
- Jan '26 Paper accepted as an oral at the WACV 2026 Physical Retail AI Workshop
- Oct '25 Our team's NordAssist agents featured in a GitHub Universe 2025 session (TechTarget)
- Aug '25 Moderated Generative AI panel at AI4 2025
- Apr '25 Invited speaker, USF Data Science Speaker Series
- Aug '24 Panelist at AI4 2024, Computer Vision & GenAI
- Aug '23 Panelist at AI4 2023, Advanced NLP
// service
Workshop Reviewer
NeurIPS · ACL · EMNLP · ICLR · COLM
Judge
CODiE Awards (SIIA) · Conrad Challenge
Open Source
Microsoft AutoGen (bug fix) ·
Model Context Protocol Servers (docs)