I work at the intersection of AI product development, research, and cognitive science, figuring out how AI behaves, what good looks like for a given use case, and how to turn what we learn into better product experiences.

​

Since joining AI21 Labs in 2020, I’ve built and continue to lead a research team working across a wide range of AI applications, including language models, RAG pipelines, chatbots, and complex agentic workflows. We work from product goals and user needs, translating product requirements and hypotheses into clear success criteria and ways to measure progress. By studying how systems perform in real use cases, we help teams understand capability gaps, uncover failure modes, assess trade-offs, and decide what to improve next.

​

This approach combines rich qualitative data from human interactions with quantitative analysis, so we can get to the bottom of specific behaviors and still identify patterns that generalize. Bringing these methods together in a single workflow has given us a shared foundation of data, metrics, and evaluation methods that supports product discovery, prioritization, iteration, and more reliable automated evals.

​

My background is in Cognitive Psychology (Psycholinguistics). I have a PhD, completed five postdoctoral fellowships, and spent many years doing research and teaching in academia. For over fifteen years, I studied how people formulate, express, and interpret complex thoughts. That work still shapes how I approach AI: looking closely at behavior, being careful about what the data can tell us, and building practical methods that help teams make better product decisions.