Mixed-methods researcher and AI data program lead. At Apple I run the human data programs behind AI models: what gets collected, how quality gets judged, and how those judgments reach the teams training the model.

AI/ML data programs, four of them run end to end
designed and run end to end, from field research to executive readouts
gesture-recognition accuracy gain, from ground-truth studies I designed and ran
Where I’ve worked
Now · frontier AI
Apple AI/ML Data Operations Team
Public interest, research & media
Pathcheck Foundation · Ideas in Action · DePaul University · Project NANDA · TEDxBeaconStreet
Enterprise & consulting
Deloitte · P&G · L’Oréal · Colgate Palmolive · Starbucks · Adidas · Samsung · Oracle · Shell · Canon
And what I work with
Research areas
Speech and audio · Sensor and motion · Vision and UI understanding · Multilingual text · Wearables and on-device · Accessibility · Human factors
Methods
Contextual inquiry · Think-aloud and diary studies · In-depth interviews · Usability testing · Surveys at scale · Regression, conjoint, MaxDiff · Log and A/B analysis · Quantitative and mixed-methods analysis · Human-in-the-loop evaluation · Ground-truth studies · Annotation and labelling design
Tools
Dovetail · UserTesting · Qualtrics · Optimal Workshop · SQL · Python · R · Tableau · Figma · Miro · Jira
Programs and delivery
Program design and staffing · Vendor and partner management · Budget and unit economics · Quality gates and acceptance criteria · Roadmaps and stakeholder readouts · Research operations
What I research
The questions I get hired to answer
Explainability and intent
How people decide when an AI’s output can be trusted, and what an interface has to show for them to act on it.
Human-in-the-loop evaluation
Study design that puts real users’ judgments into the training and evaluation loop, at the pace ML teams ship.
Bias and fairness
Surfacing algorithmic risk in AI/ML data pipelines and turning it into research practice that teams follow.
Point of view
Three bets I’m making about people and AI
01
Aim beats speed
Prototypes now cost hours, so the expensive mistake is building the wrong thing well. The decisive work moves earlier: which problem is real, and what evidence would settle it. That is research, on the critical path.
02
Master evaluation, master the model
Base models are converging. The advantage left is knowing whether yours works for the people it was built for. That puts human evaluation on the roadmap: who judges, by what criteria, and how their verdict reaches the team.
03
Trust is earned in the handoff
Adoption turns on one moment: when control passes from person to agent and back. People need to know what the agent will do, when it hands control back, and how to correct it. Accuracy cannot save a bad handoff.
Selected work
How it actually went
UX Research at Apple AI/ML Programs
15% gesture-recognition accuracy gain
Mixed-methods research on how people use voice, gesture, and on-device AI, shaping human evaluation and study design for Apple’s AI/ML systems.
Apple AI/ML Programs
Human-AI Interaction
Mixed-Methods Research
Sage Spark AI: Decision-Making Simplified
55% faster task completion in usability testing
Usability testing on an AI ideation tool that helps founders validate business ideas fast
Emerging Tech UX
Agile User Research
Intent-Driven Design
Human AI Interaction
MIT Pathcheck: Amber Alert for Health
57% Reduction in Public Health Misinformation
Decentralized AI for Public Health Crisis Response | Hackathon Runner-Up
HealthTech
Crisis Response
Rapid User Interviews
Concept Testing
In their words
Speaking & media
The findings travel
I speak about how people trust AI, explainability, and the research that keeps models honest.
Figma x SmashingConf, New York 2025
· Speaker, Intent-Driven AI & Explainability
UXPALOOZA 2025
· Featured Speaker, Designing with Intent
Kumbhathon Startup Festival 2025
· Invited Keynote, Intent-Driven UX for AI Products
UXPA International 2024
· Presenter, AI & UX: Future-Proofing Public Health
GenAI @DePaul 2024
· Panelist, Generative AI & Pedagogy Faculty Roundtable
Project NANDA, MIT Media Lab
· SF Bay Area Lead, the human half of agent trust









