 I don't have a presentation so I'm just going to speak for my notes, which is right here. I'm the co-founder of a mobile ad startup in Chennai called Zestans. Now we're called Pogli Mobile. I would say that I'm a beginner in big data. That's a disclaimer, right? There's an awesome use case that Siddharth, I think, now jerks it through again. And we're not at that scale and I wish we were, but... So we are really beginner. I'm going to share some of our experience quickly in five minutes and then possibly move on to the panel for the discussion. So why do we use big data? We use big data for real-time analytics, right? Near real-time analytics. I wouldn't call that real-time. It's definitely not real-time. Near real-time data analysis. We're also beginning to use it for data modeling. We're trying to model data and trying to kind of build probability models and trying to find out what user might click next. We're trying to figure that out. We're also now beginning to use this on a new project where we're trying to compute the, you know, the CPM of the ads. If somebody is advertising, you probably understand that at real-time. So we're doing some real-time bidding, we're connecting to a few exchanges such as Google and NextAge and Pogmatic and so forth. And we need to come up with a number that we can tell them, hey, this is the price that we can give you for this ad. And if they like the price, they will sell our ad. If they don't like it, they won't. So we're also using a lot of data analysis. I wouldn't call it big data analysis. We're definitely doing a lot of data analysis on the fly. We're not trying to come up with that number. So these are the three use cases, right? Database, modeling, and data analysis for bid computation. We are currently doing about 20 billion impressions per month approximately, right? So that's a decent number. We started with the regular, you know, I'm a technologist myself. In big data, unfortunately, I haven't made too much hands-on because I think we're better than me in the company. You know, they sort of have taken over. So we started with MySQL, right? And we scaled it up reasonably well. So we were using MySQL as early as when we were doing about 300 to 400 million impressions a day. But it was becoming quite difficult to manage because MySQL, as Navjul pointed out, it's built for old TV. It's not built for raw analysis. It was a long use case. We knew MySQL, so we went with that. Along the line, we realized that, hey, this is not what MySQL was built for, right? So let's figure out a better system that can do this analysis. And we evaluated a bunch of technologies and we came across Hadoop. And Hadoop nearly fit the bill for our analytic system. So we straight away adopted that. It is not as easy as it sounds. It took us a while, a lot of struggle and a lot of time on to it. And it's now a fairly stable system. It's become very mature as well. So mySQL pointed out that the whole time it has to take three days. It doesn't take so much of time. But it probably takes half an hour or one hour maximum. And if there's downtime, we can literally bring it back. And even uptime, it's reasonably high. So the systems have become more mature right now. So we started with SQL. We moved on with big data technologies like Hadoop. We do have, I think, a petabyte, nearly a petabyte scale cluster. Nearly. I'm not sure if it is multiple petabytes. I know it's somewhere there. It's generating a significant amount of data every single day. And it's getting higher in every quarter. The data velocity is very, very high. The growth is significant. And so we're trying new approaches. So one of the new things that we're doing is quickly, we are adopting Headspace. And instead of writing, simple things, simple changes. So Headspace gives us a different kind of capability. You can kind of read really fast the data. It does intermuse CFS. And instead of writing streamed data, for example, we're aggregating something like what he's done. So we are pre-aggregating the data encounters. So instead of storing, you know, it's a simple technique, right? Instead of storing Microsoft Internet Explorer, we see it one time. And then some end up later, you could do a simple thing, like represent Microsoft Internet Explorer with a byte value, which is just 0, 1, and just increment the counter by 1 when you see that, right? So a lot of simple things like that is what we've been trying to on Headspace and we're trying to make the system better and better. That's our experience about big data. I would say we are in the initial stages. You know, it is a reasonably big cluster, 100 plus machines. I think nearly 100 plus machines and growing steadily. I think we'll hit the 100 million mark very, very soon each month. So, I think it's very likely. Let's get there. All right. And so anyway, I think big data is a great opportunity. I think we all just, we need to get our feet wet as startups in Chennai. Of course, companies like Paypal and Amazon are already using it significantly well. And I think there's a lot to learn. So I'm here actually not to teach you, but I'm here to learn. And I think we all can have a great discussion and try to figure out what is big data, where we use, what are the opportunities, how to apply it, and so forth. And that's why we organize this whole thing. All right. That's about my talk. Thank you.