Building the dashboard for real-time glucose monitoring

Why I Joined
I first came across Levels while researching whether we could integrate with it at Zero. Glucose management and CGMs were completely new to me, and I was fascinated by the idea of designing a tool that gives people real-time insight into their own data with such immediate cause and effect. Eat a heavy meal — you see it. Didn't sleep well — it's there. Had a few drinks — expect a spike. When I found out their design team was just two people and there'd be room to work on real problems, I had to join.
The Problem
The product worked. The problem was everything around it. In the U.S., a CGM requires a doctor's prescription. So to use Levels, you had to sign up, get approved by an in-house physician or join a research trial, wait for the device to ship, download the hardware manufacturer's app, connect the sensor, and then connect to Levels. All before you ever saw your first glucose reading. On top of that, the hardware margins were thin and the devices weren't cheap. Levels had raised $38M at a $300M valuation, but the core challenge was finding a profitable way to put CGMs in the hands of healthy people in a regulatory environment designed for people who were sick.
The Biggest Shipped Project
The biggest shipped project was redesigning how glucose data was presented. Every competitor did the same thing — show the current reading as a number and give you a chart of the day. But numbers only communicate so much, and while that might work for power users, we wanted to reach more novice users. I redesigned the home screen around a visual ring that translated blood sugar readings into ranges — in range or out of range, with colors signaling which. It made it possible to understand today's readings without knowing or caring what the specific numbers meant. Stability over time instead of a single data point. Beyond the app, Levels partnered with Truepill to offer members bloodwork — the missing piece alongside real-time glucose. CGM gives you one data point continuously; bloodwork gives you dozens at a snapshot. We designed the results experience to assess markers against your demographic and health goals, visualize what's in and out of range, and deliver semi-personalized video walkthroughs explaining your results. I also worked on their AI food logging feature — a calorie detector that used your phone's camera.
Running Design Solo
About a year in, Levels went through a significant pivot and laid off a large portion of the company, including both other designers. From that point on I ran the entire design function solo — mobile, web, marketing, internal tools. The company reorganized so that each engineer owned their project end to end — finding a problem, validating it, sketching solutions, designing, building, launching. I worked alongside front-end engineers on the projects where UX and UI mattered most, but the structure made it hard to work efficiently and stay focused on the right problems. The culture was genuinely different from anywhere I'd worked. Memos over meetings, async-first, building in public, low ego. Engineers designed in Figma, designers pushed commits. The company operated more like a research lab than a startup.
Looking Back
If this role had come a couple of years later, it might have ended differently. At one point, our CEO wanted me to essentially become a design engineer — working in the same process as the engineers, just with an emphasis on more design-heavy tasks. I wasn't comfortable enough pushing code at that point. We only had GPT-3. With the tools that exist now, that hybrid role would've been a much better fit.