There is a particular kind of failure that haunts always-on devices, and it almost never shows up in the lab. The firmware runs fine for a day, a week, a month, and then somewhere out in the field it quietly stops. No crash log, no stack trace, just a device
How we're building a dog facial recognition system from scratch using landmark alignment, metric learning, and a custom-trained embedding model. The problem EverBowl is Hoomanely's precision healthcare device for dogs. It monitors eating and drinking habits, ocular temperature, sounds, and behavioral patterns over time, turning routine
A pet-health monitor doesn't get to reboot when it feels like it. It sits on a kitchen floor, watches a bowl, weighs every meal, listens for chewing, runs computer-vision models on-device, and uploads insights, continuously, for months, unattended. The Linux underneath all of that is not a detail.
Why gating and physical safeguards matter before firmware ever gets involved A feature that turns on at the wrong time isn't just a firmware bug. In real embedded products, accidental activation can create electrical stress, false sensor readings, unexpected power draw, startup instability, or confusing factory-test failures. Sometimes
Hardware redesigns rarely stay contained to one subsystem. A change in one place ripples outward, and what looks like a simple reduction in sensor count can cascade into new structural constraints, new joining challenges, and a fresh round of stress calculations. This is the story of moving our ring scale
Every engineering team runs into the same wall eventually. A product starts with a clean codebase, a small team, and a clear vision. Features go in fast, decisions get made pragmatically, and development moves quickly. Then growth happens: more developers join, more requirements land, more integrations get bolted on. Changes
A dog's weight is arguably the single most predictive number in preventive care. A slow upward creep is the earliest signal of obesity; a slow decline can be the first whisper of kidney disease, dental pain, or a dozen other conditions a vet would love to catch months
Every embedded engineer has experienced it. A prototype is running perfectly on the bench. Sensors are streaming data, processors are executing complex algorithms, peripherals are synchronized, and logs indicate everything is healthy. Then someone accidentally unplugs the power cable. The board shuts down instantly. Power is restored a few seconds
Presence detection sounds solved—motion sensors have shipped for decades. But a real living-space sensor must spot a person sitting dead still, ignore furniture, work in the dark, and not confuse the dog for a human. Our first trigger, a single-channel time-of-flight lidar, failed nearly all of these. Here'
Modern device systems don't usually fail because one request is too large. They fail because many reasonable requests arrive at the wrong time, under the wrong retry behavior, with no shared contract between client and server. That pattern shows up most clearly during reconnects. Devices go offline for