At Somo AI, I was in charge of leading prototyping, which basically meant I got to build from napkin to product. That included taking ideas from rough sketches all the way through software, hardware, and enclosures, usually within one to two-week timeframes, to figure out what would actually work in our smart hotel system.

The goal was to create a true smart system that didn't rely on Wi-Fi or crowded 2.4 GHz and 5 GHz channels. Instead, we used an automotive CAN bus, so the system didn't need internet access or cell service, you could literally be in the middle of the ocean and it would still work. Everything talked to each other: lights, climate control, shades, sensors like presence detectors and door sensors, and more. It was a fully integrated, end-to-end system.

One of the coolest parts was that these weren't just experiments that sat on a shelf, they were used in investor demos. Often, the only products investors ever saw were these prototypes, neatly packaged in production quality enclosures. They helped Somo raise over $20 million in funding and generate $1.2 million in revenue.

There were countless tight deadlines where I had to prove whether an idea was viable, which forced me to get really good at building proof-of-concepts that worked on the first try, and when they didn't, I got equally good at cutting traces and patching boards to make them work fast.

I worked with a ridiculous variety of technologies: real-time operating systems, multiple microcontroller platforms (ESP32, Raspberry Pi Pico, STM32), and every kind of sensor you can imagine, PIR, thermal, current sensing, millimeter-wave radar, IMUs, and more. I also worked with motor control systems, industrial protocols like RS-485 for DMX lighting over CAN bus, and RF communications ranging from sub-GHz to Bluetooth Low Energy.

The entire system was designed to retrofit into existing hotels without tearing anything apart, while still being smart enough to optimize both energy usage and guest experience.