Lab
Ideas & explorations
Short-form thinking. Not polished articles — raw ideas, things I'm testing, and beliefs I hold.
Belief
The best products bridge worlds. The gap between a $5 Arduino and a $50,000 MATLAB license is where the most interesting work happens.
Exploring
TinyML on ESP32-CAM
Getting TensorFlow Lite models to run on 2MB flash. The constraints are the interesting part.
Idea
What if hardware development took days, not months?
GenAI coding agents that understand hardware constraints. We're prototyping this at MathWorks and the early results are compelling.
Thought
Community over solo
Running Chennai Makerspace taught me more about product thinking than any course. When you watch people struggle with your tools, you learn fast.
Exploring
Audio keyword spotting
Arduino Nano 33 BLE + edge inference. Small model, real-time audio classification without the cloud.
Belief
Failed experiments count. Half my best product instincts came from side projects that went nowhere.
Exploring
Sensor fusion for small robots
Combining IMU + ultrasonic + camera data on a microcontroller. The math is simple, the real-time constraints are not.
Idea
Developer content at 2x speed
Using GenAI to draft hardware tutorials and getting-started guides. We piloted this and cut production time by ~50%.
Signal
Power optimization for always-on devices
The next generation of IoT needs devices that run for years on a coin cell. Sleep modes, duty cycling, and wake-on-event architectures.
Thought
Embedded is eating software
Every device is getting a brain. The engineers who understand both the hardware constraints and the AI possibilities will define the next decade.
Signal
Hardware partnerships are product strategy
STM, ARM, Arduino, Raspberry Pi — the ecosystem play matters more than any individual feature. I've seen this firsthand managing these partnerships.
Exploring
Computer vision on cheap hardware
Cameras + microcontrollers + object detection. The ESP32-CAM costs $6. What can we do with that?