Seattle, Washington
Research scientist working at the intersection of computer science, behavioral science, and economics to build agentic AI systems and understand how those systems are reshaping knowledge work.
How people really use AI systems at work, the value they get from them, and how the shape of work changes with agentic systems in the loop. How we build agentic infrastructure to augment human productivity.
Key paper: AI in the Enterprise: How People Use M365 Copilot Chat
Building novel measures from behavioral trace data using LLM-based and other inference techniques and over population scale data: macroeconomic indicators, health signals, workflows.
Key paper: Abstracting Cross-Domain Action Sequences into Interpretable Workflows
Turning those measures into forecasts and policy-relevant findings — unemployment and migration from internet data, the occupational reach of generative AI, and who is included in or left out of the digital economy.
Key paper: AI and the Democratization of Knowledge Work
Research lead · Microsoft Copilot division
A multidisciplinary applied research team on the future of human–AI productivity, combining computer science, behavioral science, and economics. I own organizational vision, research strategy, talent development, cross-company partnerships, and the translation of research into next-generation productivity and AI experiences.
User behavior measurement for the AI era
An analytics platform that characterizes what users are actually doing inside AI systems. I led the development of inferred measurement to improve AI chat systems, and owned both the research agenda and the product improvement pipeline.
Hyperlocal air quality sensing for cities
A hardware-to-analytics platform for city-scale air quality: state-of-the-art sensing hardware paired with a cloud data stack for ML-based calibration and analytics. I oversaw data and ML engineering, managed public and private partnerships, and led overall design, deployment, and research.
Economic and financial forecasting from internet data
An economic modeling and forecasting system built on internet data that drove an internal trading portfolio of more than $100M, plus a Microsoft product providing forecasts for major macroeconomic data releases. I led data modeling and engineering and served as liaison to the Treasury trading team.
Google Scholar, August 2026 · 10 granted U.S. patents
More than 75 peer-reviewed papers, with a best paper award (CSCW), honorable mentions (CHI), and multiple test-of-time awards (Web Science, ICWSM). Full list on Google Scholar.
Ten granted patents spanning language-model methods, search interfaces, location inference, and mobile social systems.
LLM-based inferencing; machine learning across feature creation, clustering, time series, and model selection and evaluation; exploratory data analysis; data visualization; experimental statistics; R and Python. Working knowledge of NLP (sentiment, embeddings) and deep learning (PyTorch).
User behavior modeling, agentic systems development, productivity measurement, macroeconomic modeling, and measure development from naturalistic data.
Strategic visioning and research agenda development with high business impact, people management, and leadership of multidisciplinary teams.
I am glad to hear from researchers, collaborators, and teams working on measurement, AI, and the future of knowledge work.
[email protected]