Would it look good here? A lamp on a product page says very little about how it will light up your living room. So people hesitate, or buy the wrong one.
Try it at home first. In the Hue app you now point your iPhone at the room. LiDAR maps the floor and walls, and a Hue lamp appears at its real size, glowing, right where you would put it. Twelve lamps to try, from table lamps to ceiling lights.
From "would this fit?" to "I can see it", in one tap.
My part. I worked on the Hue app in Flutter, on this augmented reality experience that brings the lights into your home before you buy them. And I worked on the cloud behind the app, which has a chapter of its own below.
Inside an app people already trust. It lives in the app people use every day to run their lights, next to their own rooms and scenes, on the same cloud that keeps those lights within reach when nobody is home.
The cloud behind every light switch
A short trip, or a long one. At home, the Hue app talks to the bridge over Wi-Fi, and the bridge passes the command on to the lamps, which form a Zigbee mesh together. Simple, as long as you are home.
Ask a voice assistant to turn on the lights, or switch them off from the other side of the world, and the same command travels much further: through the assistant's cloud, through the Hue cloud, and back into your home to the bridge.
What the cloud does. It decides which apps may talk to which home, keeps track of the bridges in every home, and keeps them healthy with firmware updates and diagnostics.
It began as a single application on Google App Engine and grew, piece by piece, into some forty microservices on Google Kubernetes Engine.
Shipping without surprises. Everything that runs is written down in one GitOps repository, with a branch for each environment, and Flux pulls it into the clusters. Nothing reaches production without a pull request two people approve, with a diff of what changed inside each Docker image.
A whole cluster can be built from code with one Pulumi command, so the cloud now runs in several clusters around the world. Alexa expects an answer within a set number of milliseconds, so the cloud has to be close to the people asking it. And when one cluster fails, the others keep the lights within reach.
Seeing what happens. More than 100,000 custom metrics flow into Datadog dashboards, so the team can see what every service is doing before anyone at home notices something is wrong.
If the cloud goes down, every bridge in the world goes with it.