Curio — AI Tactile Display
A refreshable tactile tablet that lets blind users read diagrams. I work on the ML and computer-vision layer — edge-detection and segmentation pipelines that turn complex visuals into something fingertips can read.
Turning machine-learning research into things people actually hold, tap and trust. Computer vision for the fingertips of blind readers. Language models that tutor like a patient teacher. Edge AI that stays calm under real-world noise.
Say hello
My work lives at the edges — where computer vision meets a blind user's fingertips, where a language model becomes a patient tutor, where an edge device makes a split-second call in a noisy factory.
I'm less interested in benchmarks than in whether the thing actually helps the person holding it. If a model can't earn its place in someone's day, it hasn't shipped yet.
Completed a Masters in AI & Machine Learning at Coimbatore Institute of Technology and now pursuing an MBA in Human Resources at Vellore Institute of Technology — because the hardest part of AI is almost always the people around it.
AI products, research prototypes and client work — most of it shipped at dVerse Technologies, some built independently or at hackathons.
A refreshable tactile tablet that lets blind users read diagrams. I work on the ML and computer-vision layer — edge-detection and segmentation pipelines that turn complex visuals into something fingertips can read.
A conversational tutor built for school students and competitive-exam aspirants — daily quizzes, audio recaps, live AI tutor and a question-paper view. I work on the AI layer and the learning-assessment logic.
An AI studio that maps rug designs to yarn colours already in stock. Cuts colour-matching time for one of India's most respected handmade-rug houses.
Product & AI engineering work delivered as part of the dVerse client portfolio — one of several industry collaborations shaping how AI shows up in day-to-day operations.
Every project starts from a person, not a dataset. The system comes after.
Before writing a line of code, I map the real-world need — who's underserved, what's broken, what already works.
Literature, prior art, open-source models. If the wheel is already round, I don't reinvent it — I improve the car.
Smallest working version first. Measure, iterate, kill what doesn't pull its weight.
Software only matters when people use it. I ship, I listen, the next version comes back sharper.
What I take on — for consulting, research collaborations and open-source work.
Research to production — building models that actually run in the real world.
Designing for users who are typically left out — tactile interfaces, assistive AI, inclusive UX.
Conversational learning, personalised assessments and AI workshops — I've trained 300+ students and 150+ teachers.
Email is the fastest way to reach me — I reply within a day. Open to collaborations, internships, research partnerships and ambitious hackathons.