Umyal Dixit — Gurugram, India
I'm Umyal. I'm a CS student, but what I'm really studying is the space between human minds and intelligent systems — because I believe the most important design problem of our time isn't how to make AI more powerful. It's how to make it understand a human.
I grew up thinking differently about problems. Not asking 'how do I solve this?' but 'why does this exist, and what does the world look like once it's gone?' That instinct led me into AI, into emotion research, into hardware, into blockchain — not because I was chasing trends, but because each of those fields held a piece of an answer I was looking for.
I build things that sit at the edge of what's technically possible and what's humanly meaningful. I research emotion classification in machines. I'm building a system to prove that photographs are real in a world drowning in fakes. I host events, speak in public, lead teams, and stay up too late thinking about things that don't have names yet.
26-class emotion space
Timeline
First hardware project ships
Built a sun-tracking floating solar panel from scratch — C++, Arduino, servos, sensors. Realized hardware + software together is where I want to live.
Wound sensor deployed
Computer vision system for medical wound analysis. First time I built something with genuine diagnostic stakes.
Emotion AI research begins
26-class facial emotion classification model. Most ML work stays at 6 basic emotions. I wanted to go deeper, to the hard ones humans struggle to name.
Veris + Sukku in parallel
Deepfake detection at the hardware layer. A proactive AI companion that notices. Two of the most ambitious things I've attempted — running simultaneously.
The Questions I'm Trying To Answer
Can a machine recognize the difference between a smile that means happiness and a smile that means pain?
Can we build AI companions that people actually trust — not because they're programmed to seem trustworthy, but because they've earned it through consistent emotional intelligence?
In a world where any image can be faked, what does it mean to prove something is real?
How do we build systems that don't just respond to humans — but adapt to them, remember them, and grow with them?
What I Can Build
Deep learning, model optimization, computer vision, NLP, emotion modeling — the full AI pipeline from research to deployment.
From Arduino circuits to React frontends to blockchain verification layers — I build across every level of a system.
I don't just engineer features — I think about how people feel when they use something. The UX instinct underneath all my technical work.
I can stand in front of a room and make it understand why an idea matters. That's rarer than any technical skill.
Stack