█████╗ ██████╗ █████╗ ██╗ ██╗ ██╗ ██╗ █████╗ ██╗ ██╗██╗ ██╔══██╗██╔══██╗██╔══██╗██║ ██║ ██║ ██╔╝██╔══██╗██║ ██║██║ ███████║██████╔╝███████║██║ ██║ █████╔╝ ███████║██║ ██║██║ ██╔══██║██╔══██╗██╔══██║╚██╗ ██╔╝ ██╔═██╗ ██╔══██║██║ ██║██║ ██║ ██║██║ ██║██║ ██║ ╚████╔╝ ██║ ██╗██║ ██║╚██████╔╝███████╗ ╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝ ╚═══╝ ╚═╝ ╚═╝╚═╝ ╚═╝ ╚═════╝ ╚══════╝
Duke ECE undergraduate. Founder of Project Zūl, MPact Glove & Project Pivot. International robotics champion (FIRST · Robotex · MakeX), Forbes India contributing writer, and researcher building perception pipelines at the seam of computer vision, robotics, and applied ML.
Three things I've founded — a STEM-education initiative and a paired hardware + software training system for boxing.
A STEM-education initiative empowering 200+ students across schools in Jammu & Kashmir with hands-on robotics, electronics, and programming — through a structured six-lesson curriculum. Approved by Hon'ble Lt. Governor Manoj Sinha.
A sensor-driven boxing training glove using high-fidelity accelerometers and gyroscopes to measure punch force, velocity, and frequency in real time. An onboard ESP32 streams to a companion mobile app; sensor fusion delivers actionable insights on technique, power generation, and performance over time.
A computer-vision mobile app for boxers without access to formal coaching. Real-time motion tracking and pose estimation deliver personalized feedback on punch accuracy, footwork, and defensive form — 94% pose accuracy at <40 ms latency. Coachless training across 70+ partners. Patent pending · Play Store-ready.
A short trail of internships, research, and reporting — each in a different problem domain.
Researched and wrote in-depth pieces on India's role in the global AI annotation economy and the use of digital twin cities for infrastructure planning. Featured in Forbes India's News and Upfront sections.
Engineered a CUDA-accelerated workflow that automated exploratory data analysis across 4.5 TB of robotic sensor data, generating 45 visualizations and cutting analysis time by 90% (6 h → <30 min) using CUDA + Numba.
before │█████████████████│ 360m after │█▍ │ 28m
Accepted by Curieux Journal. Applied LightGBM and Gaussian Process Regression to structured nuclear data from the Grand Accélérateur National d'Ions Lourds to predict electric octupole transition probabilities and low-lying negative parity excitations.
institution: Duke University degree: B.S.E., Electrical & Computer Engineering expected: May 2030 focus: Computer Vision · Machine Learning · Robotics coursework: Signals & Systems · Linear Systems · Probability · Embedded Systems · ML Theory prior: Dhirubhai Ambani International School, Mumbai ib: 42/42 — Math AA HL · Economics HL · Physics HL · Spanish B SL · Chemistry SL · English SL igcse: 11 A*s / 11 — board examinations honors: International robotics champion — represented India at FIRST, Robotex & MakeX led: TEDxYOUTH@DAIS (2025) · Your Money Matters (financial literacy club) · Team Viraasat
A few places Project Zūl has been written about.
The fastest channels. Open to summer 2026 internships in computer vision, robotics, and applied ML.