Media Week NYC theCUBE + NYSE Wired | Anand Pradhan, ICE

Clips
News
More from Media Week NYC theCUBE + NYSE Wired

Anand Pradhan

Head of AICoE & Mortgage Data

ICE

AI trends shaping the future: The role of data engineering and scalable platforms in industry transformation

Artificial intelligence is a game-changing technology in the enterprise world, increasingly adopted for its potential to enhance human capabilities. AI trends are driving industrial transformation across sectors.Large language models, infrastructure optimization, generative AI and platforms for scaling have already hit the road when it comes to notable AI trends, according to Anand Pradhan (pictured), senior director of technology at the AI Center of Excellence at Intercontinental Exchange — he also runs the NYSE Launchpad Lab

play_circle_outline Top three AI trends: large language models, infrastructure optimization, platforms for scaling.

play_circle_outline Entrepreneurial opportunities in fast-paced venture development with AI technologies.

play_circle_outline Data engineering being crucial for organizing and utilizing data in AI applications.

play_circle_outline Coolest thing seen: AI decision engine successfully making decisions.

Info
Transcript

Anand Pradhan, ICE

Anand Pradhan

Head of AICoE & Mortgage Data ICE

At the NYSE, Anand, the head of the AI Center of Excellence for ICE, talks about the top areas of technology movement he sees. Companies streamlining data for large language models, infrastructure optimization, and platforms for scaling are key. Companies like Oklo and JetCool are solving energy problems. The trend towards entrepreneurship is strong, with accelerated venture development and opportunities for bootstrapping. The intersection of business and technology in New York is enabling innovation and problem-solving. ICE focuses on streamlining operations... Read more

explore Keep Exploring

What are some key aspects in the development of large language models and generative AI products? add

What are some ways that companies are able to quickly launch products without the need for traditional venture capital funding? add

What advantages did companies who were already organizing their data ahead of time have when the AI wave happened? add

What did they attempt once with different AI models in order to make a decision? add

bolt Powered by CUBE AI

Anand Pradhan, ICE

search

>> Hello, welcome back to theCube Coverage here at the NYSE. I'm John Furrier, your host of theCube. Of course, we're here at the NYSE New York Stock Exchange floor. It's theCube's New East Coast Studio in collaboration with NYSE Wired community. Anand this year, who's the heads the AI Center of Excellence for ICE. Anand, great to see you. We had lunch the other day with some big time customers. I can't state their name. Very confidential. Great to have you back. You're like a regular guest now on theCube, you're like an analyst for us. You see a lot of people, but you're also involved in a lot of the tech and security involved here at ICE, Parent Company of NYSE. You're in a very unique position. You're building tech, you're a practitioner, you're managing and securing in your entire operation, but you're also talking to vendors all the time for many reasons. One, they're going public on the New York Stock Exchange, but also you're looking sourcing opportunities to be a customer. So you get the best of both worlds, kind of a celebrity.

Anand Pradhan

>> I love the access to this company and the idea and that gives us also idea to build something new. So all good. All good.

>> So you see what's hot out there. I know you talked to hundreds of companies a week. Sometimes you get in-depth briefings and they share some secret sauce with you. I know you sign NDAs, you really can't go into great detail on some of the companies, but if you could share the categorical top three areas you see activity where there's needle moving opportunities, what would be those three areas?

Anand Pradhan

>> To my knowledge and basically where I see the technology movement that happens. Anytime you see people talk about the large language model and generative AI and what they see as the product. But what I notice is that how these products are being built, so the data that goes into the large language model. So there are very interesting companies who are trying to streamline the data, taking many sources of data and getting streamlined pipelines. I met a company yesterday, it just blew my mind what they're doing in terms of data. And they're providing data to the large language model trainers or basically OpenAI or Lama, Meta and things like that. So that's one aspect. The other aspect is all the infrastructure that it needed to be built. So there are layers of companies who are working on that space, optimizing hardware, memory, chipset, storage. There are many coverage who are there. I love that space because invariably any improvement that happens, it helps in general for training or any other processing. So that's a positive side effect that I see, next generation of compute and everything will be evolving. The last thing is anybody who has a platform to scale, those are the companies who are going to, at the end of the day, going to benefit a lot because of the large language models will keep changing or keep it creating. We have seen a huge change in the last one year, but all these things. And one of the other things I can tell you that Oklo is a company which got listed here, which is solving the power problem by building these nuclear reactors, the small-scale nuclear reactor, which are localized for powering the data center.

>> Which company was that?

Anand Pradhan

>> Oklo.

>> Okay, cool. We saw JetCool.

Anand Pradhan

>> JetCool.

>> They've got innovative solution out of MIT.

Anand Pradhan

>> Correct. So basically those kinds of things. Cooling solutions to energy-solving solutions to network devices, storage, you name it. So that fascinates me more than what's out there on the internet.

>> And I got to say, getting to know you, you're a very curious person, but you're very entrepreneurial, I can see it in your eyes. And one of the things I share with you from a value standpoint is that it really is a great time to be an entrepreneur right now. If you look at the opportunities, you're seeing some real accelerated venture development across the world. India, here in the US, now we have a global landscape, the UN's here. This world's going to change really fast because you can get up to speed with critical escape velocity very quickly right now if you can get a needle-moving opportunity to take that lever and go straight vertical up like a hockey stick right out of the gate without a lot of capital. So if you hit the formula right with AI and some of the technologies coming together, there is a confluence of these trends where if the stars align and you don't have to go through the normal process to go through venture capital, you can just hit it right out of the gate and hit a home run.

Anand Pradhan

>> Yeah. And many companies are doing that bootstrapping with nothing. Literally a couple of developers and you can put something out there kind of thing. I see many of the companies starting two or three months, they have a product out there. So that's fascinating. There are many tools and platform coming together to build, let's say website or an application. And all these, the large language model providers have made APIs available so that's fascinating. Like OpenAI is at the top. So many companies are evolving like that. So yeah, I see a fascinating time and a movement. We'll see what happens in the next year, but I have never seen the technology movement like .

>> I was talking to a friend the other day and there was an old expression, you go back a couple decades, a couple cycles ago, and it stood around for a while. It was called, "Oh that entrepreneur, he's throwing spaghetti against the wall to see what sticks." It's an old expression because if you throw spaghetti, it sticks to the wall or it doesn't, either it's ready or it's not. And so it's also a metaphor for you trying too many things. I don't yet know what my product market fit is. And that was a shotgun approach except the risk was so high because you can't do all those things. You can't quote boil the ocean over as they say in startup nomenclature. But now you can. You can actually throw a lot of spaghetti at the wall right now with agents, with technology because you can stop start very quickly now. I know you and I have chatted briefly about this, but this is now a new flywheel that is because the data's accelerating so fast into the market, the arbitrage opportunity of data, the acceleration and enablement of data is creating a new dynamic, and we're seeing just so many examples of this. What's your reaction to that and do you have anything you want to add to that? Because I think this is a new trend we're seeing, which is it's an entrepreneurial equation that's different than we've seen before. Net New. There's other iterate, fast go, lower cost to get a server that's cloud, but now with the data you can move even faster.

Anand Pradhan

>> Yeah, and the core of it, the ability to solve problem or look at the problems and have a solution in mind. Now is the time where the implementation has been fast-tracked. But the way I see it, there is a problem you need to identify in the solution. Everything else has become easier these day.

>> But copycat too, you can look at someone else and say, "Oh, I don't need to be first, I just need to be second and faster." That's the other thing.

Anand Pradhan

>> Yeah. But one other thing I would suggest is that understanding a little bit of details of the technology because pretty quickly the cost and the maintenance becomes over it. So keeping an eye on how it can be done to scale, that's what we are doing. We understand all the lower-level detail before we get into something. So we are taking the platform-first approach to make sure-

>> The ICE platform?

Anand Pradhan

>> Yeah. So basically we want to make sure that all the benchmarking is there, all the ground truth for any of the things are established before we roll out anything. So the testing system is super important and that's another opportunity for many companies to build a testing platform.

>> Let's talk about ICE for a second. How would you describe your environment? Obviously your infrastructure is pretty rock solid. You guys got good security practice. I don't want to go there. It's pretty much good. So check. But what are you enabling for AI? As you look at AI opportunities, you're seeing new ways to innovate operationally, increase your security posture, but also generate new value. We heard productivity gains from Glean was in here. They were on the show floor, so there's logo lighting up on the big board here and the NYSE. The founder there was doing enterprise search, walked in, now he's got this awesome enablement. The productivity is enabling companies to finding stuff to move the needle just in their jobs. What do you see ICE doing that's going to be enablement for your people who work there but also your customers?

Anand Pradhan

>> Look, ICE is a global financial organization. So we have data services, exchanges, clearing assets and mortgage technology. And what we do is optimize, remove the manual labor and create value out of it. That's has been one of the core concepts we have at lower cost and how we create profitability out of that. With AI, that is even better these days. So we are focused on streamlining all these initiatives here. But what we are basically setting up is how do we take all this feature and provide it to customers? Behind the scenes we are building that platform so that we can scale rapidly experiment. And my job is to create a collaborative environment where ideas can come up from across the company and then we provide a solution, we build and we scale. So that's what we are doing at ICE and a lot of new things will be pretty soon be announced as we go into the next year.

>> How's your center of excellence going? You guys feel good about some of the cross-pollination and best practices being developed. You guys feel good with you about your experiments and your approach to AI?

Anand Pradhan

>> Yeah, going pretty well. People ask what the Center of Excellence, those kind of things.

>> It's excellent. It's the center of all the action.

Anand Pradhan

>> No, our idea is that look, you need to understand what it takes to build an AI system. Unlike the typical AI influencers on the internet, they don't talk about all the details that goes into building an AI pipeline. So we do all the right thing. I have been saying that understanding little bit details of systems and what it takes to build AI system. And so we do that thing.

>> It's a lot of practitioner work that you're putting in. I think to highlight your point, just if you don't mind me, pausing there because I think it's worth expanding is that you have to be a practitioner now to understand how it works. It's like playing chess, if you don't know what Checkmate looks like, you got to know the game you're in, and you got to take some time to get some workable projects to see some end-to-end action.

Anand Pradhan

>> You touched the right point. And we have been doing that and many people reach out to me, how do we learn AI, how do we get in to AI? And the chief scientist who built the DBRX model at Databricks, he told me, we were sitting together and he said, he tells people that, "You just build something, just start, get started and have a benchmarking solution." Those were two insightful input and we have been doing that at scale. Identify use cases, train the people, upscale as you go and set up the benchmarking for yourselves for each of the things so that we can scale.

>> Yeah, what's interesting here, what I love about coming to New York, as the intersection of business and technology, we had some folks in town from the Bay Area. The CEO, co-founder of Roboflow, they do computer vision but very developer focused, the biggest libraries out there. He's the younger generation, you're in the middle of a very big organization in ICE, but the game is still the same. They're looking at it, they're engineering, they're not just old school system guys like us who know distributed computing. We are living in the biggest renaissance of digital distributed computing, systems architecture, software and large scale problem solving. That attracts anyone who's smart, wants to be curious, wants to solve hard problems. AI is just the application software of that and or combination of the system architecture. So I think this is where I think where you guys are getting it right, is that you guys looking at it from what's the architecture infrastructure capability. And then what's the software environment going to look like that's novel for us making application change?

Anand Pradhan

>> Correct.

>> Gen AI is an application feature. The system scale is a infrastructure feature.

Anand Pradhan

>> Correct.

>> It's old school, hardware, middleware, app. It's the same stack. It's just got different slices and dices. Explain your thoughts on this. I think you agree, but this is important to simplify and then look at it.

Anand Pradhan

>> Correct. That's correct. And any application you build, you have the hardware and you have the software and you have the requirement and you test system and things like that. It's a little more complicated with AI and Gen AI because the data is heavily involved. So it becomes really more complex and it is expensive. So to run each of the iteration, keeping track of each of the iteration and benchmarking. And whoever is doing it right in a more streamlined way, they're going to win eventually without a huge cost. Other thing that I can think of is that it'll produce a lot more data and there will be another industry to keep track of the data.

>> It's so funny, data's in the middleware. I always said that for years, Dave and I always riff on this. If you look at the action, the data business outside of the folks who specialize in data, certainly exchanges have a big data focus, but even then they were built on monolithic old school legacy systems. But outside in the classic enterprise, it was a department and area of an analytics. You had a warehouse, it was all good. We validated, it's like crown jewels are in there, don't touch it. Queries, get a report. Who cares where it's late. There's no real time dashboard, but in comes data science. Okay, great. But now you've got data engineering going on. I think one distinction in this platform engineering transition to setting the table for Gen AI is that there's real data engineering being discussed. Meaning not just databases, but databases in context to the system. This is new. Not new, but only the big guys did that. Like the banks who make money on trades and ICE.

Anand Pradhan

>> Yeah, so look, many of the companies, when the AI wave happened, they started getting into the space of how do we organize our data to use it? And whoever had organized or started organizing much ahead of time, they're right there, right up there to release something. Spotify, which is listed here, when they're one of the-

>> They use Datagray there. Datagray they have recommendation engines, personalization. I like the new virtual DJ. I saw that. Maybe I'm in the beta. I don't know if anyone got that, but that was pretty cool.

Anand Pradhan

>> Yeah. Spotify, they have one of the best data tech stack, which is there for research. And I have done extensive research on that and read their technical documents and they have done something right. And so what I'm getting to is that that data organization has become a more focused point right now, access to the data. So the industry right now, whoever has traditional and antiquated data solution are thinking like a data lake and Apache Iceberg and all these different companies.

>> As the Glean guy pointed out, yeah, Spotify is a little bit different use case. They have a lot of data, not a lot of legacy systems, but enterprise have different systems, different identity access. So the Spotify is a great example that enterprises want to get to. And companies like Glean, they're working on that. What's the coolest thing you've seen in the past couple of weeks? If you pick one conversation, you can name the company or not name the company. You can just categorically talk about it. What is the coolest thing you've seen in the past couple of weeks?

Anand Pradhan

>> Coolest thing always is if somebody can build a decision engine using AI. I have seen something where literally a decision can be made using AI. That to me is super cool. I understand what it takes to build that. So that is-

>> You saw the level of difficulty?

Anand Pradhan

>> Yeah.

>> And to get that wasn't a fake demo. It was legit.

Anand Pradhan

>> Yeah. We have attempted once. So basically we understand what it takes to make that, not that we are going to use that or roll out that, but we did a full-blown experiment to put different AI models together to make a decision and that's something super cool.

>> Anand, great to see you. We're at time here. We got to run. We're going to the 11th hour. The exchange is empty. If I keep going, I could compete with mad money down there and go, just keep going. We'll have a marathon. We'll have a bake-off. Who can go longer? Great to see you.

Anand Pradhan

>> Good to see you too.

>> Thanks. Really pleasure working with you.

Anand Pradhan

>> Thanks.

>> Okay. Thank you for watching. I'm John Furrier here at the NYSE, the new location for the NYSE Wired community with theCUBE coming together, building out our East Coast Hub here. Thanks for watching.