 So let's say you have a graph where it's a million nodes. Like for example, let's say you're Twitter. I don't know how many people have joined Twitter, but I'm sure it's more than a million. How many people have joined Twitter? They have 50 million, I don't know. Yeah, in 2009 it was about a million, and then when Obama joined and Oprah did it just went straight up. Right. So over a hundred million? I think they last time they went public, they said they had 80 million. Okay, so we've got enough data points. It's pushing a hundred million in pattern matching. So you've got these giant graphs, right? So a lot of what you want to do on those graphs is you want to find patterns in this graph. So you want to find, for example, you want to find and give me an instance of, like to tell me who is connected to both Barack Obama and also Lady Gaga, right? So you want to pay a list of people who have connected to both, or who has been retweeted by both of those people. That'd be an interesting query to ask. So in order to sort of issue that, so that type of query turns out to be sort of a sub-graph pattern matching query where you have a graph of who's retweeting whom or who's connected to whom. And you want to sort of find some part of that graph in the larger context of much larger graph. So it turns out that you can do it pretty easily on a single node. When your graph is split across hundreds of nodes or even thousands of nodes, like, for example, Facebook or Twitter, then it becomes much more complex. And so we have some research trying to solve that problem. That's very interesting. And I'll talk a little bit more about your role as a data scientist, chief scientist. I said data scientist before, chief scientist at a depth. Different data scientist. Although you're dealing with data, so. It's true, it's true. I mean, we are a data company and we build a product for data scientists, but yeah, but you know, I mean, so essentially what happened was, so the Hadoop DB project at Yale was sort of, we were working on it right when Hadoop was sort of getting started, at least starting to get a little bit more popular. So by 2008, well actually really early 2009, we published a paper that got a lot of traction in the industry. So we sort of showed how to combine Hadoop, which obviously by 2009 was really taking off with data-based systems. So Hadoop is well known to be very good for unstructured data and for sort of, for multi-structured data, whatever you want to call it. But on the other hand, it's also known for being not so great relational data. If you're a relational data in Hadoop, you can do it, you can query it with Hive and it works, but it's not so optimized for that. And it's actually received a lot of criticism for not being optimized for that. So today, what a lot of people do is they have Hadoop and a database system sitting side by side. And they have these connectors between them that sort of ship data back and forth between Hadoop and the database system. So I think that's nice. I think that solves certain problems, but I don't think it's a good long-term solution. It's really having two systems, both of them are parallel systems that sort of do scalable data processing. It's just not, it's sort of, why have two when you can have one? So the goal of the HadoopDB project was to try and create one system. So make Hadoop, fix its problems for structured data so that therefore it can be as good a database as for structured data. So that, we published a paper about that in 2009 and we, and it got a lot of traction. So eventually venture capitalists started coming to us saying you should really commercialize the technology. There wasn't, although I definitely did want to sell a company at some point in my career after seeing Snowback all those times, but it was still a bit early for me to do it. I was sort of, I wasn't really expecting to sell a company so soon, but given how much excitement there was in the industry and the pressure from VCs, I took the jump. We started a company, right towards the end of 2000, actually in the middle of 2010, I think it was, we actually officially started it. Which VCs went in to that one? So the early round VCs were sort of local boutiquettes VCs and New Havens were a VC called Launch Capital. And also the VC, sort of the state of Connecticut has a VC firm called Connecticut Innovations and they also were in on that first round. Of course now I'm sure you know, a couple months ago we raised a bunch more money at $8 million from Bessemer and Norwest, which are actually, 8 million, well actually, so 9.5. It's sort of complicated, official filing was 9.5 million dollars, but Bessemer and Norwest themselves put $8 million into the company. So. The partner from Norwest. Matt Howard. Well, I'm not sure if he's a partner, but he's the guy who's working with us, I'm not sure what exactly status is there. And at Bessemer it's Felder Hardiman, who's also Felder Hardiman, who's the guy behind Vertica. So obviously, there's some comfort level there from the Vertica days. So yeah, so anyway, so over time we managed to raise a bunch of money and so now the goal is to actually build a product that has an impact in the industry. So that's sort of the history there. So my role, I think your question was, watch my role there. So my role as Chief Scientist is to really sort of, the most important thing is to help detect transfer. So to bring the idea from the lab, we have, you filed three patents, it's about to file a fourth patent, that came out of my lab at Yale with all these double chats and all my students at Yale, trying to sort of move that over to the company to make sure that- Does Yale get a piece of the action? Yale does, Yale owns a poly company. Yeah, yeah, absolutely. I mean they own the patents, but to any work that I do at Yale, the patents that are filed are owned by Yale. So Yale definitely got a piece of the action there. So anyway, but yeah, so the first thing I do is tech transfer and also sort of, I participate in the development meetings and sort of help with some high-level design points in the product as well. And natural as a founder, you also have various business stuff you have to do as well. I wouldn't say that's Chief Scientist stuff, but it's still random stuff that you got to do when you start companies. How do you like that? Is that a welcome change? Or is it like, yeah, I really want to do this, but I have to because I'm- I mean, it's a, I'm definitely learning a lot. You know, I mean, a lot of this is new to me. I learned a little bit from Bredaker. Especially the earlier days of Bredaker, I got to see how it was done, but now I'm actually doing it myself than seeing other people do it. So a lot of it's new to me. But I mean, for now it's fun. I find because it's new, I'm learning a lot. Like it's still, like it's intellectually challenging. We'll see if it still stays interesting to me over the long term, but yeah. But for now it's, for now, I'm having a good time. Excellent. All right, Daniel Abadi from Yale and Hedapt. Appreciate you coming on theCUBE. Great insights. Congratulations on all your success. Love that story. Tech geek, professor, Yale owns the patents, building out companies. Congratulations. Hey, keep on knocking them, man. You got Hot Street going. Yeah, keep stepping up to the plate and building the companies. Yeah, cool. All right. Thank you very much. Thanks.