 from San Francisco, it's theCUBE, covering Spark Summit 2017, brought to you by Databricks. Welcome back to theCUBE, and we are pleased to have our first guest here today. He is a customer of Databricks, and also doing some exciting things with Spark. So welcome, Nathan Murath, our Senior Software Development Manager from Autodesk. Welcome, Nathan. Thank you. Are you happy to be here? Absolutely, very exciting. Is this your first Spark Summit? It is, absolutely, yep. First time here, first time at a Spark Summit. A lot of fun, a lot of people, a lot of energy, so I'm very happy to be here. Well, before we dive into some of the exciting things you're doing with Spark, maybe tell me what you were hoping to learn at this summit. I think, I'm really interested in learning what's coming next. Autodesk is a technology company. We build products, we build software, and we're always looking at the future figuring out what we can build and what we can leverage this amazing technology for in our own tools that we then offer to our customers. And did you just attend the keynote? I did. What did you think? What stood out to you? A lot of interesting things that I want to go home and try, basically, or take back to the office and try, is a lot of these things are very applicable to what we're doing on a day-to-day basis. We've also got George Gilbert on the show, and George, we're going to dig in a little bit. Maybe you have a question for Nathan about what he's doing with... Yeah, Nathan, for those of us who are anti-Diluvian, in other words, having been born before the big flood that flooded Nozark, what was... Tell us about the types of products that Autodesk builds and how Spark helps people who use those tools. Sure, so Autodesk is a company we do a lot of different things, right? Autodesk primarily builds software for the design and make space in three or four different verticals and disciplines. One is media and entertainment, one is manufacturing, one is architecture, engineering, and construction. The group that I'm a part of and the software that our team is responsible for building is mainly around the cloud and mobile products for the architecture, engineering, and construction industry. Specifically, we have a suite of products that we brand BIM 360 that basically are tailored to the construction industry and various personas and various steps, depending on where you are in the life cycle, the building, a vertical structure, a bridge, a hospital, a stadium, and we provide software for those individuals. Can you tell us a little bit about that life cycle and then the life cycle of a project like that and where Spark can help a customer who's thinking 360 and not a particular product? Absolutely, so the life cycle is pretty complex, but it starts usually on the computer like this, where you'll design your building or your structure, whatever it is, heavy 3D graphics, that's where Autodesk, the company started. From that point on, you do a number of coordination exercises with the various disciplines in a construction project. So your architecture, your structural, your plumbing, your heat ventilation, air conditioning, people that are all specific disciplines, then you actually go on site and you start building this structure, whatever it may be. And when you build the building process that typically could take multiple months or multiple years, depending on how large the structure is, is when people are going to start leveraging some of our tools even more. Typically, some of the things that we see a lot of is when you're managing a construction site, you will see on a daily basis, hundreds, probably if not thousands, of construction issues. You know, this sheet of glass here is broken, the drywall fell off, this beam is going through a wall, you name it. Construction site problems happen every day. You want to know if the beam is going through the wall. Exactly, absolutely. So typically that generates a lot of data. And to the point where our customers can possibly feel overwhelmed by their own data because there's just so many things that get generated in our systems. So this sounds, it actually sounds like where IT operations is the discipline of I've got all this infrastructure and I'm getting all these alerts when little things go wrong and I don't know where that necessarily the root cause is or what I should attack first. Is that sort of what your... It's where we're going. Yeah, absolutely. Right now we're tackling, we're starting, we're still early stages with kind of the machine learning, data science applications to the products that we do and build, but right now what we're tackling, we're just trying to help our customers gain insights on their own data. So when you're swimming in this vast ocean of data and you don't know where to start or typically a construction site the size of a stadium or a campus or a huge office building, you don't know where to start typically. So what does this vast pool of data look like and how specifically are you using Spark to help make sense out of it or prioritize what you should look at? So a lot of what we're doing now is we're using image data and text data. So what happens is your superintendents when they walk around a construction site to figure out what's going on, what's broken, what's working, what should I focus on today? They will walk around with our mobile devices or their mobile devices using our mobile software and take pictures and write descriptions of things that they see walking around the construction site. So they've generated hundreds of thousands of these construction issues and where we're leveraging Spark is to help build classification models on top of that data, be it image and textual data to help bring to the surface and bring to the top the things that are most critical to our customers. Typically one example is on a construction site any problem that's related to water is usually considered a big problem. So if we can help among hundreds of thousands of issues that happen on construction site kind of identify what, hey, Mr. Superintendent, you have these 10 problems that are related to water, whatever they may be, you should probably focus on those first. And that's what we're leveraging. Things like Spark, technologies, machine learning, data science to build the products. And are you learning from all the customers who use the product or in other words, do you need their data to get smarter or is it rules that you're building? So right now we're working with a subset of our customers through which we've gone into a number of agreements where they were okay with us working with them very closely to possibly use some of their data that was generated in our tools and systems to help build our model. So we're absolutely not looking at the entire data set per se. Did you see anything in the keynote with the structured streaming that's now down to a millisecond which is truly impressive for Spark or in the deep learning that might simplify traditional machine learning? Did you see anything there that looks like it might have an impact on the type of app you could build? Very much so. So I think all the streaming applications are very relevant because more and more on the construction sites or more and more construction sites are being censored with whether it's webcams, cameras, temperature detection, dust detection, air quality detection, exactly, IoT all over the place so when we can start collecting the data from those devices and streaming into our systems we can more proactively notify, warn our customers, people on site, either security risk, any danger situation or simply this is happening right now in your construction site you might have to wake up because it's the middle of the night and go check out what's going on. This is actually of great interest to us because one of our themes now where customers are telling us that they're trying to figure out what type of analytics especially like the machine learning training would happen in the cloud and what type of analytics would happen on site or on customer premise. Are you doing the training up in the Autodesk cloud and then are you doing, would the models be evaluating and executing sort of on site close to where the data is being captured? So right now again, you know, early stages so a lot of those questions we're still trying to figure out and understand what's best, what's best for our customers, what's obviously the most secure and things like that. A lot of the training that we're doing today is in the Autodesk cloud so we use a lot of our cloud infrastructure where the data resides for our products, of course, to build and train our models, essentially. Well, we only have a couple of minutes left but I wanted to dig into maybe some of the lessons learned. You said it was early days so some things you could share with the community here on theCUBE that would help them. Somebody maybe just getting started with Spark and some of the valuable lessons you've learned you'd want to share. I would say probably get started now is probably my piece of advice. I think we're all going in this direction and a lot of technology and it's interesting because even the construction space that I'm in is maybe not considered the highest tech discipline or industry which makes sense and is obvious but even in the construction space we're going in the direction of kind of using machine learning, data science, Spark-like technology. So I would say get started now. That would be my piece of advice because there's a lot to learn and things move really fast. Okay, so if you can complete this sentence, Spark has finally enabled Autodesk to blank and start with Spark. I'm trying to get a soundbite out of here. Yeah, I think so. Spark has finally allowed Autodesk to build valuable customer services facing machine learning and data science products for our customers. And then the business outcomes for that are being closely watched by executive sponsors or how's that working? They are, but again, early days, right? Any large corporation like Autodesk, early days is a lot of moving parts so we're still feeling the waters of it now. All right, George, last question goes to you. I guess from what you've seen today and anything you've heard about also coming down the roadmap, how might you expand the application that you are building in terms of thinking new possibilities, pushing the boundary? I think so, internet of things is something that we're looking at, right? And I can very well foresee being part of this solution and ecosystem as well as just allowing, I think allowing our customers to push and pull their data into our systems to leverage our technologies or to pull it back out to plug it into their BI tools or things like that. And I think that's something at least for our enterprise customers will be very valuable. All right, well, Nathan Murath from Autodesk, thank you so much for spending time here on theCUBE. We're going to have to get back to the show. It looks like the show floor is open now so get out and network with some of those 3000 attendees. Perfect, thank you very much. Thank you so much. And thank you for watching. We'll be back soon with more guests here on theCUBE.