PhD Done, Hello Stanford!

I’ve completed my PhD at the University of Michigan and started a new chapter as a T32 postdoctoral fellow at Stanford’s SNAPL (Systems Neuroscience and Pain Lab), mentored by Sean Mackey.

Wrapping Up the PhD

My dissertation, Usable and Ubiquitous Privacy-Aware Sensing Devices, started from a simple observation: it doesn’t matter how amazing a technology is if people won’t use it. And people won’t use it if they don’t trust it. The most effective fall detector is a camera, but my grandma would never put one in her bathroom—where most elderly people actually fall.

So I spent my PhD designing sensors that guarantee privacy at the hardware level—not through policies or promises, but through physical constraints: microphones that physically can’t capture speech, cameras that strip out personally identifiable information before the data ever leaves the device, wearables that monitor health without becoming surveillance tools. When the sensor itself enforces the privacy guarantee, you can finally deploy these systems in places that were previously off-limits—like bathrooms, or bedrooms, or anywhere people actually need help.

Of course, constraining what the sensor can see means you have to get creative with the machine learning. It’s easy to build systems that work when you have tons of privacy-invasive data. The real challenge is making it work with what’s left.

That tension—between what you need to sense and what you’re allowed to sense—drove most of the technical work. It created a true hardware-software co-design cycle: figure out what the ML could do with constrained data, revise the sensor to capture better signals, then revise the ML to take advantage of the new sensor.

Fellowships from Meta and Rackham gave me the freedom to pursue these ideas, and I was able to publish the work at venues like CHI, MobiCom, IMWUT, and PETS. Most of all, I’m grateful to my advisor Alanson Sample for giving me the space to pursue ideas that seemed risky at the time, and to my committee, collaborators, and the UMich CSE community for pushing me to make the work better.

Why Stanford and Pain?

The pivot might seem large, but I see it as an evolution of the same core values. During my PhD I became increasingly interested in how sensing can address real clinical needs—particularly for conditions like chronic pain that are hard to objectively measure.

Pain management hits all the challenges I care about: privacy, because pain is intimate and stigmatized; fairness, because pain is systematically undertreated in certain populations; and usability, because tools must work in real clinical settings. It’s also deeply universal—almost everyone knows someone who suffers from chronic pain. I know this personally: years ago, a car accident left me with a spinal injury, and it fundamentally changed how I think about what it means to live with pain. There’s something powerful about building the tools you wish you had.

I’m also pursuing an M.S. in Epidemiology alongside my postdoc, and it’s reshaping how I think about research questions. The trust principle holds here too—clinicians won’t adopt technology they don’t trust any more than patients will. But clinical trust isn’t earned with the shiniest new AI model. It comes from asking the right questions upfront: not “what’s the most accurate model?” but “what’s the right outcome to predict?” and “how do we validate that we’re actually measuring what we think we’re measuring?”

If my PhD taught me anything, it’s that the best technology is the kind people actually trust enough to use. That principle becomes even more critical when people’s health and lives are on the line.