About
Computer scientist interested in what happens when elegant ideas meet hard constraints in hardware, software and AI
My Story
Hi, my name is Abhi. I come from a beautiful coastal city in India called Vizag. If you had asked ten-year-old me what I wanted to be when I grew up, I probably would have said a writer. As much as I love storytelling though, I guess it wasn't meant to be in the end. Because a year later, I discovered programming and realized I could build the things I imagined instead of simply writing about them. I couldn't help but dive deeper into understanding how software actually worked. I spent the next few years teaching myself how to code across different languages and tech stacks, and the more software I built, the more I realized there was an entire world beneath it that I didn't understand.
The software was starting to make sense, but the devices running it still felt mysterious. So rather than taking the most direct path into computer science, I chose to study Electronics and Communication Engineering for my bachelor's to better understand the hardware behind the abstractions, before pursuing a Master's in Computer Science at the University of Florida to deepen my understanding of AI and software systems. Today, I'm most interested in building intelligent software at the intersection of AI, systems, and hardware; technology that's deeply engineered under the hood, yet feels simple and almost invisible to the people using it. And as with any obsessive interest, the deeper I go, the more questions I have.
Some Questions I Keep Coming Back To
- How do we make frontier AI accessible to everyone, not just those with access to massive compute?
- How do we make inference dramatically faster and cheaper without sacrificing capability?
- How do we build models that are deeply personalizable, real-time, and capable of running entirely on-device?
- How do we build AI with taste? AI that creates with intention instead of imitation?
Some of these questions came from the work itself.
Selected Experience
UF SCAN Lab
At the SCAN Lab, I work on automating complex PCB design and modification workflows as well as 2.5D semiconductor package design. My magnum opus is a production pipeline that can modify schematics and PCB layouts, place new components, and automatically route the resulting nets while producing deterministic designs that pass electrical and design-rule checks. One of my main considerations is building algorithms that are robust enough to work across boards with wildly different geometries and constraints. So far, my work has contributed to an award-winning research paper published at the 2026 IEEE Electronic Components and Technology Conference (ECTC).
Defence Research and Development Organisation of India (DRDO)
During my internship at DRDO, I worked on compressing and deploying neural networks on ARM Cortex-M microcontrollers, achieving a 4.1× inference speedup and a 6× reduction in model size while improving accuracy. Working within the constraints of embedded hardware gave me an early appreciation for extracting maximum capability from limited resources; a mindset that still shapes how I think about efficient inference and on-device AI today.
Teaching
As a teaching assistant for Programming Language Principles at the University of Florida, I helped students navigate everything from parsing and type systems to interpreters and compiler design. I particularly enjoyed the challenge of explaining difficult ideas without hand-waving away the details. Because if you can't explain something clearly, there's a good chance you don't understand it as well as you think you do.
What I Enjoy Building
Software Engineering
I like taking hard problems and turning them into software that actually holds up outside the environment it was built in. I enjoy thinking through architecture, edge cases, and abstractions—and doing the less glamorous work it takes to turn something that works once into something you can actually rely on.
AI & Machine Learning
I'm interested in AI across the whole stack: building models, building the systems that make them fast and practical, and building the applications people actually use. Inference, personalization, multimodality, and squeezing more capability out of limited compute are the problems I keep coming back to.
Products & Experiences
At the end of the stack, I still care about what someone actually gets to do with the thing. I like building products that are useful, thoughtful, and occasionally a little magical; from a computational photography app to tools that help people create, learn, and get things done.
Low Level Systems
I'm increasingly drawn to the layers underneath the applications we build: compilers, runtimes, parallel computing, accelerators, and the hardware itself. I want to understand how decisions across those layers shape performance, and how designing them together can push beyond what's possible at any one layer alone.
Beyond Computing
When I'm not coding or researching, I spend a lot of time reading, mentoring other students, and pursuing whatever happens to catch my curiosity. Over the years, I've also had the opportunity to take on a number of leadership roles, both during undergrad and at UF. I love photography, learning languages, cooking, spending time outdoors, and occasionally trying my hand at music. Fitness is also a big part of my life, I lift regularly and am working toward becoming a well-rounded hybrid athlete, balancing strength with endurance.
Honestly, I can go down just about any rabbit hole. One day I might be reading about the Roman Empire, trying to understand economic equilibrium the next day, thinking about the philosophical and ethical implications of AI after that, and then dropping everything because Veritasium published a new video. I'm especially drawn to connecting disparate ideas from different fields.
If you'd like to see how these ideas become software, take a look at some of the projects I've built.
Explore Projects