Varun Cumbamangalam
Engineer. Architect.Full-stack problem solver.
I build connected products from the device up: embedded hardware, firmware, edge ML, mobile apps, and cloud infrastructure. I have spent seven years working across healthcare, wearables, and fitness tech in four countries.
Close to the hardware
Sensors, firmware, BLE, and edge compute built for the physical world.
AI under real constraints
Models shaped around latency, devices, users, and deployment.
Across the full stack
Mobile apps, APIs, cloud systems, and the data paths between them.
From prototype to production
Engineering decisions that turn early ideas into dependable products.
Products I designed, built and shipped.
On-device OCR
Vakalatnama.pdf
Matters in today's causelist
0
VakeelBrain
Hearings this week
Shipped for
Selected projects
View all
01 · Backend · Frontend · AI/ML
VakeelOS
Legal practice software for Indian advocates, with court workflows and AI-assisted drafting.
India-first Workflow & Residency
Case study
02 · Backend · Frontend · Mobile
ArogyaLens
One multilingual platform for patients, doctors, and hospital teams.
13 Languages Supported
Case study
03 · Backend · Mobile · Frontend
ArogyaSync Platform
Clinical data sync from bedside vitals to an audit trail.
99.9% API Uptime
Case study
04 · AI/ML · Hardware
On-Device Activity Recognition
On-device activity classification on a 64 KB RAM wristband.
23+ Model Iterations
Case studyWhat I build
From the first sketch to production, across hardware, apps and the cloud.
Brand & Identity
Marks, palettes and design systems that carry across product, site and pitch.
Websites
Fast, animated marketing sites that are live and earning enquiries. Pick one to see it scroll.
Mobile Apps
Flutter and native Android apps, offline-first, with BLE where the hardware needs it.
IoT & Hardware
Devices, firmware and the sensors in between, from breath analysers to wristbands.
AI & ML
Models trained, quantised and shipped to kiosks, phones and microcontrollers.
Cloud & DevOps
Multi-cloud platforms with monitoring, fail-closed releases and cost control.
How I use AI in the build
AI tools are not shortcuts. I use them to prototype architectures, test edge cases, and explore options early. I still make the design calls and own the production code. That applies whether I am training disease detection models or building autonomous agents.