 Okay, we're back here live in Las Vegas. This is SiliconANGLE and Wikibon's the Cube, our flagship program where we go out to the events, extract a signal from the noise, go live, go wall-to-wall coverage, get all the signals, share that with you. John Furrier, the founder of SiliconANGLE, join my co-host Dave Vellante, co-founder of wikibon.org. And we're here with Greg Duffy, the CEO of Dropcam. And we talk to a lot of geeks under the hood talking about software-defined data center networking, but it's great to have a cool product on as well as talk tech. Greg, welcome to the Cube. Oh, thank you very much. And Geek, I resemble that reminder. You do, we're going to go under the hood here. But Dropcam is a real successful product. But there's some stuff going on under the hood we're going to get into. But talk about the product, the success you're having and what the company's doing. And then we'll just talk about how you wire it all together. Yeah, absolutely. So we started Dropcam to kind of answer a really simple question. What's going on at home when I'm not there? And it's actually been a crazy journey building the ecosystem of technology required to answer that simple question. So we take in more video than YouTube. We process all that video on the Amazon Cloud. We build a lot of technology to deliver just the most important moments to you and deliver it to all of your viewing devices. So that's basically what Dropcam does. Love to answer any questions you have about something. So Nannycam, you know, Cam for the business, any camera basically, just pop it in there. We should have one here. We got to get you guys to hook us up to some Dropcams here to kind of get that extra bird's eye view. But basically this is classic case of surveillance video or video where you're pulling in massive amounts of data, okay, video streams to somewhere. Take us through how that works. So do you have any hosting at all? Is it all clouds? Is it all on Amazon? How did you wire it together? So, you know, first of all, one of the first things we learned is that Dropcam is not just about surveillance. People use it in their everyday lives. Surveillance is kind of like and security are kind of secondary use cases. So we've got people capturing video of like their families, their kids, their pets. It's actually a lot of fun on top of being useful. All that video does come into the Amazon cloud where we wrote a lot of software to process, index and store it. So for example, we have a system called Nexus that stores in a custom data structure built on top of DynamoDB and S3, all that video that's coming in. We also have a custom system called activity recognition that tries to learn the different type of events that happen on your camera and deliver you alerts to only the categories of events that you really care about. So all of it's built on top of AWS but there's still quite a lot to do even after you've taken care of the basic compute and storage infrastructure. So before we get into more of the technical geek stuff, how much scale are we talking about on the video streams that you're ingesting in? Just give us a kind of a feel for the order of magnitude. It's petabytes every month and in terms like we process billions of events every year. So it's a huge scale. You get a good discount on your EC2 bill. You're not storing the video. Are you storing it or just processing on it? Are you actually storing it to S3? We do store it. So about 40% of our cameras are on a cloud video recording plan and that means that we store either seven or 30 days from each individual camera. So it takes back, it's really interesting story. I mean, a lot of people who know you know the story but tell us a story about how you started the company. Yeah, it is really hilarious. It started off with my dad who had a problem with a neighborhood dog that shall we say was fertilizing his lawn and he wanted to set up some IP cameras to capture video of whoever was letting their dog do this in his yard. And he had a ton of problems with them. To set them up, you needed to know what IP addresses were, how to set up port forwarding on your router, all kinds of geek stuff. So my co-founder and I decided, hey, what if we take all that stuff that you need to know and just encapsulate it in a cloud service so that you don't, as an average user, all you need to know is your Wi-Fi password to get it up and running. And that's what contributed to the drop cam scale of today. We decided to solve all those problems in the cloud. We didn't anticipate that it would become the largest inbound video service on the entire internet because of that, but that is what happened. So initially you had trouble getting funded, right? Yeah, it's a meaty problem. It's not like the average kind of web or app startup in San Francisco. So we were doing hardware, we did all this massive video storage. People weren't sure that we could do it, but we built it piece by piece. In fact, the first version of drop cam was just a cloud service with an off-the-shelf camera that we reverse engineered to work with our cloud service that we didn't have to build hardware. And funny enough, that first camera we bought from Amazon and shipped to the customer. So it is kind of an Amazon story all the way around. Okay, and then you purpose built your own camera. That's right. Okay, and you were funded initially by Mitch Kapoor, is that right? Mitch Kapoor started off with our seed round and a few other angel investors. And now that the idea has kind of taken off, we brought on Excel, Menlo, IVP and Kleiner Perkins and later rounds of finance. They already invested? They're already invested? Yeah, that's right. So how much did you raise? The total today was about $48 million. So it's a big chunk of change. That must have been a nice step up from the ramen noodles, rubbing nickels together, trying to figure out how to shave the bills from Amazon. So I like to tell people the first version of drop cam we actually shipped from my apartment. Every single box of the first thousands of cameras that went out, we packed up by hand after buying them from Amazon and then we reshipped them out to the customer. But now I'm hopefully never going to have to touch packing peanuts again. We still carry cameras around. So there's hope buddy. So I mean, Dave and I having the same experience with our CrowdChat app that we spun out as a separate company. It looks really like a simple app on paper but no one really taking it seriously. But if you look at how we wired it, it's a similar story. It's a lot of complexity involved in kind of wiring, the ingestion, a lot of, there's a lot of nuance. Tech involves not just, hey, throw database up and do it. So I got to ask you, when you get the cloud and let's go deeper into the tech. So what's going on in Amazon? What stack are you using? Take us through some of the, let's get, let's get Gigi using auto scaling. Just take us through the stack. Yeah, absolutely. So a few of the technologies we're using, I mentioned we're using DynamoDB and S3 as the underlying data storage for this custom video data structure that we've developed. That data structure is managed by something called Nexus and Nexus is the cloud video server software that Dropchem developed to run on EC2. Nexus is written in Scala, which is kind of a cool language choice for people who like working on cloud. It runs on the JVM, but it's a much more pleasant language than Java. So our cloud team works on that. We also have a really cool analytics system that we've just recently started using inside of Dropchem which uses elastic map reduce. So the reason we do that is there's a bunch of non-video data that comes into Dropchem. For instance, all the data coming from cameras about Wi-Fi signal strength, all the data coming from Nexus about video reception. And we collect about 300 million data points every day into unstructured data storage. EMR goes through and summarizes and structures that data so that our admin interface can help us do things like, we've all heard of AB testing websites to optimize traffic, get people to convert to different goals that you have. But we can actually do things like AB test Wi-Fi drivers. And that's because we think about Dropchem as an entire cloud ecosystem. So the black art of determining whether a Wi-Fi driver is actually better than another one, we can do that with data in the cloud. So that's another piece. Do you have search on this as well? How are you searching that DynamoDB? Solar? No, we don't actually search that data because all that in the way that you would with solar, all that data that comes in is actually unstructured video data. So what we have to do is add structure to it. So that means we had to develop a bunch of custom algorithms that all run on EC2. There's nothing to do what Dropchem does out of the box. So do you do the structure on the ingestion? We do it on the ingestion. We do a lot of processing on both the camera side and the cloud side in order to do things like activity recognition, our classification. So you're using SQS at all on this? Yeah. How are you handling all the notifications on the queuing? Yeah, so SQS, we use RabbitMQ on top of EC2 and SQS in order to do things like our activity alerts to your phone. So when we do a push notification, like we have people catch burglars all the time with Dropchem and that all happens because of a queue in the cloud that pushes an alert to your phone and suddenly you see a burglar in your garage and you call the police. We actually had someone do that up in Washington recently and they got video of the police chasing the burglar out of their house, which is pretty cool. And you can always stream it live on JustinTV or whatever live stream as well. But actually the cool thing is you can do that even just with us. If you want to make a clip public or make your camera public, it's really easy to do so, although everything is default private. But we've had streams of video like our Frankie the tortoise camera who Frankie the tortoise is a 50 pound tortoise that carries a Dropchem on his back and streams live for about eight hours every day. Some of these public cameras get hundreds of thousands of viewers every day. But since we take in so much video, it's actually quite easy to put out so much video too. So you guys using Redis at all? We don't use Redis. No, we do key value type stuff. Actually, it's kind of an interesting thing about Dropchem. Caching doesn't help that much. And when you think about it, every piece of video is new and it's going to be viewed either zero or one times. And very rarely is it going to be viewed a bunch of times except for the public camera use cases. So we have to be fast even without caching which makes the problem that much more interesting. Yeah, you got to actually go essentially wire speed almost basically camera speed. Well, great. What's next for you guys? I've got a lot of cash in the bank. What's the plan? What's the headcount look like? What are you doing this year for growth? Our engineering team is growing really fast. We're hiring aggressively there in all areas from the camera to the cloud to even our apps which are also very important for displaying real time video and getting alerts. The engineering team we're hoping to grow by two or three X over the next year. So that'll be growing our team of 50 people right now to over 100, maybe 150. And where's your office at? We're in downtown San Francisco. And we're working on all of these problems that we mentioned to do it even better from computer vision problems to video delivery, making everything really fast and nice. And we're also constantly looking for new ways that we can help people know what's going on at home. So that's what we do. Fantastic, certainly mobile phones. You just got to help the four G networks get to 5G, 6G and get faster connectivity. That's right. Got to help on the internet side of things in the US. Go call your cable companies, call your wireless companies and tell them you want more bandwidth. Oh yeah, that's going to happen soon. Well, thanks for coming on theCUBE. Really appreciate you. Go to Dropcam, great stuff. Greg Duffy here inside theCUBE. Tech geek, entrepreneur, CEO, tech athlete by our standards. Congratulations on your success, funding and continued growth. We love video. This is live theCUBE on the Amazon web services. Re-invent conference here in Las Vegas. We'll be right back after this short break.