Today’s guest post comes from Carter Maslan, CEO of Camio, a startup that converts real-time video into useful information.
I adore our dog Jane. She moved 3,000 miles away, but I still wake up to her barking and pillow-fluffing. That’s because our son shares his Camio feed with us. We get video alerts on the iPad he leaves behind as a pet monitor. These alerts come out of Camio’s AI technology, which includes hundreds of servers running neural nets competing to learn the exact moments our family wants to see. To riff on an old Google bumper sticker, “My other pet sitter is a data center” — or actually many data centers running Camio.
Camio is a cloud-based video monitoring service for watching, storing, analyzing and searching for real-time videos. It turns any camera into a smart home or business monitor, where real-time video is uploaded and saved in the cloud. Retailers can use Camio to watch and understand foot traffic in stores. Families can capture video from birthday parties and family reunions. Building owners protect offices from theft and vandalism. Care providers monitor elders living alone. Parents know at a glance when their kids get home or the dog sitter leaves.
People can use Camio to view live feeds and engage in auto-answering two-way chats for free, from anywhere in the world. They can also receive alerts when particular search conditions are met, such as when a car pulls into the driveway.
A huge technical challenge we had was to design and implement a highly scalable backend system for capturing, storing, retrieving and analyzing millions of videos — without building and managing our own servers. We wanted to spend every incremental engineering dollar on making the product great, instead of investing in infrastructure.
We had already been using Google App Engine, Google Compute Engine, Google Cloud Datastore and Google Cloud Storage, and then one of our engineers read about the Google Cloud Platform for Startups credit program. We asked one of our investors, Greylock Partners, to endorse our application to the program, which prompted us to explore Google BigQuery, Google Cloud Monitoring, Google Cloud Bigtable, Managed VMs and other services too.
As a result of being in the startup program, we were able to access valuable technical resources, including consultations with many of Google’s top engineers. The credits also allowed us to defer operating costs, so we could devote more resources to building our product.
We store videos in Cloud Storage, a high capacity, highly scalable storage solution ideal for working with extremely large data objects. The videos are processed by Compute Engine to identify patterns and events, and the metadata output from Compute Engine is indexed in Cloud Datastore, a NoSQL schemaless database for non-relational data. The videos themselves are delivered to users via App Engine. It’s all scalable, and able to grow as we grow.
There’s no advantage in our doing the things that Google does so well for us. Being maniacally focused on our product — that’s the road to success.
– Posted by Carter Maslan, CEO, Camio