Content & Model Designer
I define how AI communicates with people: its personality, how it sounds, what it shares, and how it makes you feel. I've led design teams at Meta and Huge, and now I am a model designer working on Gemini.
"When nothing is fixed and everything is fluid, guardrails and instructions are the experience."
I thrive in the in-between space, where the problem is still being defined and the stakes of getting it right are highest. That's where I do my best work, especially at 0 to 1.
I partner closely with researchers, engineers, PMs, and policy teams to get alignment on what good looks like before anyone writes a single instruction. In my experience, most model quality problems are actually design problems, and they happen upstream.
Every output is a signal. Every eval cycle is an opportunity to get sharper, together. I hold a high bar for what we ship: responses that are helpful, factual, and trustworthy aren't a nice-to-have, they're the whole point.
That's what my sweatshirt says. It also happens to describe my entire job. I'm a model designer, a title that didn't exist five years ago and still raises eyebrows at dinner parties. It sits at the intersection of UX, prompt engineering, behavioral design, and something that doesn't quite have a name yet.
This is the first in a series about designing the user experience of AI.
This work is shared selectively. Enter the password to view case studies.
Incorrect password. Try again.
I came up as a travel writer and editor, drawn to how language shapes the way people understand the world. That led to a master's in information science, then interaction design, then content strategy, and eventually to the work I do now. The throughline has always been the same: how do we build things that actually serve the people using them?
If you've worked with me, you know I want to solve the whole messy problem. Defining the operating logic of LLMs turned out to be a natural extension of everything I'd already been doing, just further upstream, with higher stakes, and a lot more interesting failure modes.
Reach out via email or LinkedIn, I'd love to hear from you.