 Live from Washington, D.C., it's theCUBE. Covering AWS Public Sector Summit. Brought to you by Amazon Web Services. Hello everyone, welcome back to theCUBE's live coverage of AWS Public Sector Summit here in our nation's capital. I'm your host, Rebecca Knight. We are joined by Ragu Raman. He is the director of FINRA, the Financial Industry Regulatory Authority. Thank you so much for coming on theCUBE. Hi, Rebecca, good afternoon. Very happy to be here. So, this is the 10th annual Public Sector Summit, I should have said. So, tell us a little bit about FINRA and what you do there. Sure, FINRA itself is the Financial Industry Regulatory Authority. We are a private sector, not-for-profit institution. Our mission is investor protection and market integrity. We are member-funded and we have a member-driven board of directors. And we engage in ensuring that all the stock market operations in the US capital markets play by the rules. So, that's the essence of who we are. And all of those stakeholders have a vested interest in making sure their rivals are also playing by the rules. Absolutely, okay. So, you're here giving a presentation on fraud detection using machine learning and artificial intelligence. That's right. What was, what were you saying? So, at FINRA, we have a very deliberate technology strategy and we constantly keep pace with technology in order to affect our business in the best possible way. We always are looking for means to get more efficient and more effective and use our funding for the best possible business value. So, to that end, we are completely in the cloud for a lot of our market regulation operations, all the applications are in the cloud. We, in fact, we were one of the early adopters of the cloud. From that perspective, all of our big data operations were fully operational in the cloud by 2016 itself. That was itself a two-year project that we started in 2014. Then, from 2016, we have been working with machine language and recently, over the past six months or so, we have been working with neural networks. So, this was an opportunity for us to share where we have been, where we are coming from, where we are going with the intent that whatever we do by way of principles can be adopted by any other enterprise. We are looking to share our journey and to encourage others to adopt technology. That's really why we do this. And I want to dig into the presentation a little bit, but can you just set the scene for our viewers about what kinds of, how big a problem fraud is with these financial institutions and how much money is on the table here? Well, I don't want to get into the actual dollar figures because each dimension of it comes up with a different aspect to it. We can say that in FINRA, we have a full case load year after year, decade after decade that end up with multiple millions of dollars worth of fines just on the civil cases alone. And then there are, of course, multi-billion dollar worth problems that we read in the media. Cases going as far back as Bernie Madoff, cases going through the different banking systems. So, there are various kinds of fraud across the different financial sectors. Of course, we are focused on the capital markets alone. We don't do anything with regard to banking or things of that nature. But even in our own case, we FINRA is composed of nearly 3,300 people and all of us engage in the full-time task of ensuring that the markets are fair for the investors and for the other participants. It's a big deal. So, in your presentation, you told the story of two of your colleagues who are facing different kinds of challenges to make your story come alive. Tell our viewers a little bit about their challenges. We spoke about Brad, who is an expert, he's an absolute wizard when it comes to market regulation. And he has been doing this for a long time. And what I shared with the members in the audience earlier today was, he can probably look at market event data and probably tell you what the broker had for breakfast. It's scary good. And we also shared the story about Jamie, who is in the member supervision division of FINRA, Wicked Smart and Extensive Experience. So, these are the kind of dedicated people that we have at FINRA. And I took up two real-life use cases, the sort of questions that they face. So, in the case of Brad, it is always the question of, hey, we are good, but how do we get better? What is the unknown, unknown there? The volume of transactions in the market keeps going up. How do we then end up with a situation where we can do effective surveillance in the market and detect the behaviors that are not of interest, that are not productive, that might be even downright manipulative? How do we make sure that we got it all, so to speak? That's Brad's thing. And that idea about these unknown unknowns, because we know we have known unknowns, but the unknown unknowns are even scarier. Exactly, they are. And we want to shed light on that for ourselves and make sure that the markets are really fair for everybody to operate in. That is where use of the latest technologies helps us get better and better at it, to reduce the number of unknown unknowns, to shed light on the entirety of market activities, and to perform effective surveillance. So that was the gist of our conversation today, how we have gotten better in the past four to five years, how machine learning-based technologies have helped us, how artificial intelligence that we started working with, specifically neural networks, have started helping us even further. Okay, and then Jamie had a problem too. In Jamie's case, member supervision, if you will, the problem is of a different context and character. There is still volumes of data, we still receive more than a million individual pieces of document every year that we work with, but in her case, the important aspect of it is that it is unstructured data. It makes sense to humans, it is in plain English, but to machines, it's really difficult. So over the past two years, we have created an entirely new text analytics platform, and that helps us parse through hundreds of thousands of different documents. Those could come from emails, it could come from Word documents, spreadsheets, even handwritten documents. We can go through all of those, extract meaningful information, automatically summarize them, even have measures of confidence that the machine will imprint upon it to say, hey, how confident am I that this is of relevance to you? It will imprint that, and then it will present to Jamie for her to use her judgment and expertise to make a final call. One thing that we are really conscious about is, we don't let algorithms completely take everything through. We always have a human. So we think of AI as really assistive intelligence, and we bring that to effect for our business. And I think that that's a real key there too for the employees, is to know that this is taking away some of their more manual, more boring tasks and actually freeing them up to do the more creative, analytical problem solving task. I think you hit that nail right on the head. All the tedious work the machine does, and then it leaves humans to do, like you said, absolutely the creative, the intuitive, and the final judgment call. I think that's a great system. How much do these solutions cost? We generally are not pricing these things individually. However, overall, one of the things that we did with the cloud was actually reduce our overall cost of technology. So from that perspective, we don't look at cost as a primary driver, although many times these things do end up costing less than the prior system that we would be in. However, the benefits that offer to our clientele, the benefit that it offers to our business, to the people that are investors in the stock market, that is tremendous, and that has a lot of value for us. So what is next for FINRA? I mean, this is a real moment for so many industries in terms of the rise of cyber threats and fraud being such a huge problem, privacy, these are the financial services industry more than, I guess maybe as equal to healthcare. This is really sensitive stuff we're talking about here. What are some of the things that you have on the horizon? What are some of the things that you're hearing from your members? So all of our members treat data security really, really special and really carefully, and we are very deliberate and very conscious about how we treat the data that is interested to us. We have two obligations. One is to treat it securely. The other is to extract appropriate insights from it because that's the other purpose of why we are being interested with the data. We take both of those dimensions very seriously. We have an entire InfoSec organization. It's composed of experts in the field. We are headed by a chief information security officer with a large team that looks at multi-layered security right from the application defending itself all the way to perimeter security. We do all of that. We have extensive identity and access management systems. We also have an extensive program to combat insider threats. So this type of multi-layered security is what helps us keep the data secure. And every day we do notice that there are additional threat factors that get exposed. So we keep ourselves on the edge in terms of working with all the vendors that we partner with in working with the latest technologies to protect our data. As an example, all of our data in the cloud is completely encrypted with high encryption. And it is encrypted both at rest and during flight. So that even in the rare case that someone has access to something is gibberish. So that's the intent of the encryption itself. So that is the extent to which we take things very seriously. I want to ask you too about the technology backlash that we're seeing so much. And you live here. So you really know about the climate that technology industries are facing. For so long, they were our national treasure and they still are considered in a lot of ways the Amazons, the Googles, the Facebooks of the world. But now we have presidential candidates calling for the breakup of big tech. And there's been a real souring in the part of the public of concerns about privacy. How, what are your thoughts? What are you seeing? What are you hearing on the ground here in DC? With specifically with regard to where we operate from in FINRA, we try not to access or use any data that is not for regulatory purpose. We are very careful about it. We don't sprawl across and crawl across social media just on a general fishing expedition. We try not to do that. All of the data that we take in, store and operate technology upon, we are entitled to use it for by policy, or by rules, or by regulation for the specific purpose of our regulatory activities. We take that very seriously. We try not to access data outside of what we have need for. And so we limit ourselves to the context. And that, if you look at, is really what the public is trying to tell us. Don't take our data and use it in ways that we did not really authorize you to do so. So the other thing is that FINRA is a not-for-profit institution. We really have absolutely no interest beyond a regulatory capability to use the data. We absolutely shut it down for any other use. We are done. That way we are very clear about what our mission is, where we use our data, why we use it, and stop. Great. Well, Raghu, thank you so much for coming on theCUBE. It's been a pleasure talking to you. Of course, thank you. Good day. Thank you. I'm Rebecca Knight. Please stay tuned for more of theCUBE's live coverage of the AWS Public Sector Summit here in Washington, DC. Stay tuned.