suriya
nameSuriya Nithyanandan
rolefounder & builder · ML researcher
locationBerkeley, CA (from Louisville, KY)
focusstartups & entrepreneurship · explainable AI research
statusbuilding something new · open to collabs
localtime
about

Hey, I'm Suriya — a CS student at Berkeley. I split my time between two things I love: building companies and doing ML research. I'm heads-down on something new right now, and on the research side I study how machine-learning models think — and where they go wrong.

On the startup side, I like going from zero — shipping the first version, getting it in front of real users, and figuring out what actually sticks. I built ProfitQuest, an app for teaching financial literacy, which led me into MoneyBot, where I was a founding engineer. These days I'm building something new (more soon).

On the research side, I work on explainable AIexplainable AIThe effort to make a model's reasoning legible — so you can see why it made a call, not just what it predicted.. Right now I'm looking at why image models mix up things that look alike — say, two breeds of dog — and using the "map" a model builds of what it sees to pin down where that confusion comes from. I just like taking models apart to understand how they actually work.

I grew up in Louisville and I'm at Berkeley for CS. If you're building something or working on interesting research, I'd love to hear about it — suriyansn5@gmail.com, or hit ⌘K.

work
Something newFounder · building in stealth
2026 — now
University of LouisvilleML Researcher · Dept. of Computer Science
2025 — now
MoneyBotFounding Engineer
2025 — 2026
projects
ProfitQuest[↗ profitquest.tech]
An app that teaches financial literacy through gamified lessons. 100+ users across 4 Kentucky schools and a Congressional App Challenge winner — it's what led to MoneyBot.
Portfolio Optimization with ML[↗ repo]
Compared investment strategies over 10+ years of stock data; a Markowitz approach hit 23.16% annual return, growing $100K to $209K.
Chronic Kidney Disease Prediction[↗ repo]
Ensemble models predicting CKD from 400 patient records at 98.75% accuracy, validated with cross-validation.
Parkinson's Detection[↗ repo]
XGBoost models reading spiral & wave drawings to detect Parkinson's — 80% accuracy at 1.0 precision.
writing

Notes on what I'm building and learning — coming soon.

books
The Almanack of Naval RavikantEric Jorgenson
On wealth, judgment, and building leverage — a lot of how I think about work.
The Psychology of MoneyMorgan Housel
A reminder that doing well with money is less about math and more about behavior.
SiddharthaHermann Hesse
A quiet reset I keep coming back to.
press
MoneyBot launches to teach financial literacyLouisville Business First · 2025
ProfitQuest wins the Congressional App ChallengeU.S. House of Representatives · 2025
Kentucky students win at the TSA National ConferenceKentucky Teacher · 2023