Faculty Spotlight: Tania Roy
Tania Roy is associate professor of electrical and computer engineering whose work focuses on developing novel hardware for AI. Her research is aimed at improving the energy efficiency of AI hardware using in-sensor and near-sensor computing. She also focuses on the development of new electronic materials that can succeed and complement silicon for future technology generations. Roy is a 2025-26 Langford Lectureship Award recipient.
Q&A
How would you describe your main area of focus, and what first got you interested in this field?
Think of today’s computers like a busy kitchen where the chef (the processor) is stuck fetching ingredients (data) from a far-away pantry (memory) every few seconds. This back-and-forth wastes time and energy — it’s why your phone’s AI assistant sometimes lags or why massive data centers guzzle electricity like small cities.
My work fixes that at the ground level: I design tiny, super-smart computer parts — called devices — that act like the brain’s neurons and synapses. They “remember” and “process” information right where it’s created, like in a camera or sensor, without needing to phone home to a big server. I use special “miracle” materials, like atom-thin sheets of graphene or molybdenum disulfide to create these parts. Imagine stacking single layers of paper to build something stronger than steel.
In short, I’m engineering the next generation of computer “sandwiches” — layers of these materials stacked on silicon — to make AI faster, greener and work anywhere, from your smart glasses to Mars rovers. It’s about turning sci-fi efficiency into everyday reality.
I have been interested in semiconductor devices since my undergraduate days. Every part of electrical engineering is interesting. But I always wanted to know what’s happening within the black boxes. When I started to get the picture where electrons were moving around in a device to create current, I was satisfied that I want to be at that level.
Recently, I see the problems of AI hardware. I strongly feel that we need to reimagine how the hardware for AI is realized to solve the issues of power and speed with AI. While other engineers are applying band-aids to solve the AI issues, we are solving it from the roots.
Could you tell us about something you’re currently working on?
We are developing smart pixels for cameras that are like the retina of the eye.
What advice would you give to new faculty members?
Being a new faculty member is hard. One should learn to face rejections. We are trained to be great researchers. We are not trained in other managerial skills. As a faculty, we are expected to be the CEO, CTO, CFO and HR of our group. Those are tough roles to fill for one person, even if the group size is small.
What has kept you at Duke?
My colleagues and students are great. The atmosphere is perfect to push boundaries of research. We have a critical mass for pushing semiconductor research.
Would you like to share something you enjoy doing for fun or relaxation?
I like to find ways to explain my research to nonexperts. That involves taking lessons in improvising and public speaking. I also love to cook for my family. My kids are very young. My favorite relaxation is watching them play.