In this episode, Mike Siswanto, Sr. Vice President at Northern Trust, explains how to eliminate the noisy neighbor problem by utilizing a total portfolio view in Snowflake, allowing for a consolidated, single source of truth for data.
In this episode, Mike Siswanto, Sr. Vice President at Northern Trust, explains how to eliminate the noisy neighbor problem by utilizing a total portfolio view in Snowflake, allowing for a consolidated, single source of truth for data.
--------
How you approach data will define what’s possible for your organization. Data engineers, data scientists, application developers, and a host of other data professionals who depend on the Snowflake Data Cloud continue to thrive thanks to a decade of technology breakthroughs. But that journey is only the beginning.
Catch up on the latest announcements from Snowflake Summit, including advancements with generative AI and LLMs, flexible programmability, application development, and much more.
Watch now here.
Steve Hamm: [00:00:00] Mike, it's great to have you on the podcast today.
Mike Siswanto: Thanks, Steve. Thanks for having me. Very excited to be on this podcast.
Steve Hamm: Now, Northern Trust is one of the largest and oldest banks in the United States, but some of our listeners might not be familiar with all of its, its businesses, the, the full scope of what it's doing.
So, could you please explain what the bank is focused on and what differentiates it from other competitors?
Mike Siswanto: So Northern Trust is 130 year old bank headquartered here in Chicago, uh, with more than 14 trillion in assets under custody and administration, and 1.4 trillion in assets under management. Uh, we are a leading provider of wealth management, asset management, and asset servicing, uh, which is the business unit that I'm.
Our clients are, uh, typically large, sophisticated institutional investors, which we typ typically classify as, uh, either asset owners or asset managers who oversee [00:01:00] global multi-asset portfolios. So an example of an asset owner is like a pension fund, endowments, uh, sovereign wealth funds, family offices, and like insurance companies.
While examples of like an asset manager is like a hedge fund. Private equity or other alternatives? Uh, so within asset servicing, uh, I work in the omnium, uh, platform technology department. And your next natural question is, What exactly is the Omnium platform technology? So the Omnium platform technology is a full-fledged middle back office platform built from the ground up to support a client's full book and records in real time with straight through processing for a.
An investment book of records, also known as an ibo, and in addition, an accounting book of records, also known as an ABO R. So what really differentiates us from our competitors in our spa is our service integrity and expertise. Our [00:02:00] primary strengths are people. And our technology, our people are what's more critical of the two As without them, the tech just doesn't work.
We are unique in the sense that our business leaders, our subject matter experts, and our tech leaders work side by side in an agile methodology. To enhance the Omnium platform and the stakeholders have transparency into the clients.
I like to think that while we are one of the biggest admins in the world, . What sets us apart is that it's all in our dna. We behave like a small boutique provider with a drive towards the highest level of service and pushing ourselves to get better. While we are also one of the biggest players on the market, what we are really proud of, is our Omnium platform, which is able to support all financial products traded in Publix and private markets, and the ability to support everything all in one.
Steve Hamm: Cool. So I, I wanna make sure I understand this [00:03:00] correctly. It seems like basically you are asset management, asset servicing that is outsource, um, Pension funds and all these, these, these big outfits basically outsource technology, data management and process to you Correct. Rather
Mike Siswanto: That that is correct.
Steve Hamm: Okay. I, I get it now.
Mike Siswanto: That's correct. Yes. So, we essentially service and administer all our clients', uh, financials, right? So whether that's, you know, uh, providing them p and l, uh, doing custody business, uh, providing 'em financial statements, reporting to the regulators, we do all that, all in a white glove service.
Steve Hamm: Yeah, yeah. That makes total sense to me. So, so that, that way they can focus on strategy, on customer relations, all those kinds of things. Yeah. Uh, okay. Very cool. Very cool. Now you've been at Northern Trust for five years, I understand. Kind of describe what exactly your role is within the [00:04:00] Omnium platform technology team, and kind of what's your career path been within the company?
Mike Siswanto: Sure. Uh, well, I play a lot of hats, uh, within the organization. I'm a senior vice president, and currently I'm the head of Onion Asset Servicing Technology. So on the delivery front, I lead a portfolio of development teams within the Omnium technology department. That's like corporate actions reconciliation, market data, data acquisition, and data and analytics.
But in addition to my delivery role, I also play a functional role as, uh, the chief architect for Northern Trust Asset Servicing business Unit with a dotted line to the chief architect of Northern.
Steve Hamm: Yeah. Yeah, it's interesting. You really are kind of a hybrid. I mean, you're a deep, deep technology guy, but you also have to understand banking really deeply as well, right?
Mike Siswanto: Yeah. Yeah. So that's, that's one of my unique skills, right. Being able to. [00:05:00] Talk at a level for all the business execs, and also being able to talk all the way down to, uh, my, uh, my development teams to be able to translate what these business requirements mean in a technical manner.
Steve Hamm: makes total sense. Now, has this been your role all these five years, or did you kind of work your way up to it?
Mike Siswanto: Uh, so it started with, you know, like the delivery teams, uh, uh, that I, that I managed, that I mentioned earlier. And then I kept on getting more and more delivery teams. And now this expanded to, you know, an overall overarching, uh, for, uh, The asset servicing, like being the chief architect to make sure that as an entire business unit we're able to, uh, architect everything in, in, uh, re resilient and, and, and re and, uh, reliable way.
Steve Hamm: gotcha. Now, before Northern Trust, you worked for Citadel, which is a very large hedge fund. What were the key management lessons you learned in your many years at [00:06:00] Citadel that you're put into use at Northern Trust now?
Mike Siswanto: So Citadel was a fast paced, high energy environment, so you had to really think fast and deliver quickly. In addition, you needed to be able to contact, switch and juggle, uh, between multiple initiatives. And I was able to take these skill sets and apply 'em to my current role today at Northern Trust. So I've been very fortunate in my career.
Thus far working with great leaders from our Northern Trust, c o Tom South to our Northern Trust, BU C of asset servicing Jeff Mac. And even to my, uh, my current, uh, boss, which who's the head of on technology, Brian Mar, all of them played an important part in my career growth. And encouraged me to, to lead and drive new initiatives that are important to Northern Trusts.
They all empowered me to make decisions, and at the same time, I have full accountability for delivery of every team and or project I lead.
Steve Hamm: [00:07:00] Good. Good. Cool. I wanna back up for just a minute here and talk about technology adoption. When and why did Northern Trust begin migrating its data and data applications to the cloud, and what's the status of that initiative now?
Mike Siswanto: The public cloud journey started roughly in 20 18, 20 19, and had different points of maturation levels throughout the organization. Like for example, uh, we have a platform called Front Office Solutions, which is our alternative investments platform, which was born in the public cloud, so it's 100% hosted and running in the public cloud.
while you have other business units, uh, within the bank starting to get a larger cloud presence now. And so therefore they're in the earlier stages of the public cloud journey. But overall, we're making a really big push to get majority of our estate, uh, onto the public cloud as we're seeing large benefits of the [00:08:00] capabilities of the the public cloud brings to our.
Steve Hamm: Yeah.
So let's, let's delve into the cloud a little bit further. Uh, when and why did Northern Trust first engage with Snowflake and the Snowflake platform, and what have the two companies accomplished together?
Mike Siswanto: So our engagement with Snowflake started roughly around Q1 2021, and during that time we started our journey to solve one of the top industry and client challenges within the asset servicing space, uh, which is essentially clients with large complex investment books in both the private and public market space.
So these are typically your very large pension funds, sovereign wealth. Family offices, et cetera, right? So when they have a very large complex, uh, private and market, uh, uh, uh, platforms, they, they tend to have to buy various, uh, vendor solutions. So for example, for on the PR. [00:09:00] Private market sites, they would have to purchase one or more vended solutions to run their privates business.
While on the public side, they would also have to purchase one or more, uh, different vended solutions to run their, their public's books, right? And so each of these softwares that you're purchasing, they, they also, uh, come with their own data. So you have to manage the data, uh, separately also. With all these vendor solutions, you're also kind of have to staff them and support them, uh, all over the place, right?
So these are all creating all these silo challenges, right? From software to data to to staff and support, right? And so what we've built and what we developed is a brand new cloud-native, cloud-based platform called Total portfolio. That we call, that we alias as T P V and the, this is knocking down these silos and essentially what total portfolio view is built on top of two industry [00:10:00] leading platforms.
So for the alternative investments, that's your private equity, uh, venture capital, uh, real estate infrastructure that's gonna be running on our front office solutions that I mentioned earlier. The public markets, which are typically your equities, fixed income, OTC derivatives, those will be running on the Omnium platform and with total portfolio view analytics.
What this enables is a chief investment officer and a chief risk officer to answer the most complex top down question. Quickly and accurately, while simultaneously the support investment and operation professionals are performing more bottoms up or asset class specific analysis. And all of this is being done through the T P V web application modules in, in the form of.
Performance exposure, risk compliance, all in near real time and accessible at their fingertips.
Steve Hamm: now, does Total Portfolio [00:11:00] view, does that run on Snowflake so that in a sense, both sides, private and public books are running on Snowflake, or help me understand the architecture there.
Mike Siswanto: Yeah, that's correct, Steve. So we, we have everything hosted in, in, in Snowflake, so that you have that, you know, consolidated, uh, single source of truth. And those modules that I, I mentioned earlier, are built on top of it, right? So some of the Snowflake capabilities we leverage, uh, for our Northern Trust platforms, uh, of Snowflake is Essent.
The ease of use of the warehouse setup, right? So they give you very, uh, you know, uh, user-friendly, ease of use knobs, for example, to set up a warehouse. You pick the size, you pick the minimum and the maximum amount of compute you want. And then you're able to pin the, those, uh, warehouses to various.
Customers and users, uh, so [00:12:00] that you are essentially eliminating a lot of the noisy neighbor problem. Right. And, and what we face is, you know, we have a very, uh, burst pattern. We have, we have a burst pattern in the sense that we have a follow the sun burst pattern as, as, uh, markets close in apac, amea, and North America.
these trades just start, get, be, are being sent from our clients to us. Right. So we had to be able to expand really quickly during those peak volume times. And Snowflake has been able to do that. Right. And furthermore, we've been able to, uh, make further use of the warehouses. By essentially, um, pinning the various data domains to each their own compute, so that like, again, uh, that you are able to do workload isolation at, at that point.
Right. Um, and then the, the, the resiliency of the storage, right? So we're, you're able to replicate across different regions, uh, [00:13:00] with, with within the, the. Public cloud provider and then time travel, you're like that's, that's, you're able to use time travel features to essentially be able to recover very quickly without having to, to be down.
Right. And, and we also take advantage of the Zero Copy Clone feature, which essentially allows us to make various intramural environments for, for various development purposes. And, and one other thing that we're, we're really exploring is Snow Park, right? So, Recently with Snow Park and the enablement of Python, this allows us to essentially bring a safe and secure facility for our data scientists to run their models on top so that they don't have to take the data in and out all the time, and they're able to just stay all in within Flake.
Steve Hamm: That's interesting. Yeah. Um, so what advice do you have for other companies or other organizations that are considering, are just starting [00:14:00] their snowflake journey or their, or their, their public cloud journey?
Mike Siswanto: I would say, you know, figure out your current constraints, whether you're on-prem or you're in, uh, or you're already in the public cloud and, and, and the data platform that you chose, right? And what are you trying to solve for, right? Is it the ability to expand compute easily? Do you have a noisy neighbor problem?
Um, do you have backup problems? Is it just disaster recovery or is it efficient data? , right. For us, one of the biggest differentiators was snowflake's, uh, data sharing capabilities. Right? And also the fact that it's cross cloud so that we're able to not force our customers to come to us, but we're able to go directly to our customers, into their public cloud, uh, provider and their region of choice, right?
And so in addition to that, you know, there's a lot of various, uh, modules and features that. Snowflake [00:15:00] has, for example, for Snow Park, as I mentioned, for the data science teams and, and snow pipe for data ingestion and even stream lit that we're starting to even explore and use that to, you know, have quick and dirty gooeys that are, are, are, um, data science team are able to even spin up themselves to, to tweak their models, uh, using user input without having, you know, me to dedicate a gooey developer to build them a a bespoke tool.
Steve Hamm: makes total sense.
now you've mentioned data sharing a couple of times.
Explain how are you using data sharing with, with some of your big, like pension fund clients and, and, and that kind of thing?
Mike Siswanto: Yeah. So we have, uh, two unique use cases, right? So. One of them is, uh, external, uh, data sharing with private listing. Right. So, so we were very fortunate to get into the private listing, uh, private preview, uh, from from Snowflake and, and you know, they've been really great partners and [00:16:00] we've been giving feedback, uh, on, on various private previews that we've been in.
And private listing is, is one of them that we've been giving 'em a lot of feedback on. Right.
Steve Hamm: well explain. Explain, yeah, explain what it is that That's cuz Cuz most people haven't seen
Mike Siswanto: it
Steve Hamm: Yeah. Go
Mike Siswanto: Yeah, understood. So, so with private listing, what this allows us to do is allows us to go to any public cloud, uh, provider at any region that Snowflake hosts and ride, uh, snowflake's backbone for essentially auto fulfillment of our data that we choose to share our data with.
Right? So traditionally, , what Snowflake has is, uh, the public marketplace where folks come in there and purchase data sets and, and or, or be able to hook up data sets, right? And, and, but like with private listing, think about it like a, a private marketplace, right? So it allows Northern Trust to essentially be able to have our own private [00:17:00] marketplace.
For our customers to be able to, uh, you know, fulfill da their data that Northern Trust is essentially, um, uh, administrating for them and sharing it with them directly. Right? So with, uh, our external sharing use case with private listing is for our total portfolio. Uh, uh, portfolio view, omnium and, and front office solutions, uh, uh, platforms that we're able to share Northern Trusts, uh, investment book of records data with our clients directly.
And in turn, what they're able to do is essentially join to the, uh, against their, their, their Northern trust data with their own data. to layer on and shape the data to however they would like. Right? And that gives them, uh, better insights in your real time. So for example, right. the clients themselves.
They can keep their proprietary alpha data within their own Snowflake account. Join that [00:18:00] with Northern trusses data to essentially generate trading ideas or hedging insights, right? So that's like the, the external data sharing use case while internally. Uh, we're, we're able to share, uh, our, our within, uh, different business units, right, with, with either the traditional data sharing or private listing capabilities.
This allows us to, essentially, as data owners, have only one single source of truth. , right. Uh, essentially emitting that to, uh, enterprise, uh, central data catalog and, and, and ha giving the capabilities to share data within the organization without actually having to copy and move things around, right? So that's very powerful.
Steve Hamm: Yeah. Now, you talked about having, um, you know, the private listing service with your clients. Is that something where you are able to monetize some of that data, or is that basically just a, a, a function of easy sharing [00:19:00] at this point?
Mike Siswanto: Yeah, very good question, Steve. So, um, I think to us, right, it, it's, um, it, it, it, it's actually a more of an operational efficiency gain, right? So if you think about it, think about traditional, uh, administration, like think about traditional asset servicing ad admins, right? They would either, they would have to call APIs, extract out data, Take that file, scan that file, ship that file, probably via SFTP to, to a client, the client would take that file and load it into their, to their, to their warehousing.
Right. So essentially you're cutting out a lot of the data orchestration and, and essentially the data delivery workflow and just streamlining it. Right? So with data sharing, that's the, that's the efficiency gain that, that, that you, we bring to our clients.
Okay. Okay. Hey, I wanna look to the future a bit here. Looking out over the next year or so, what do you think are the [00:20:00] most significant data technology trends that you see coming?
I mean, as of this recording , I believe like over time, like a lot of the data. Uh, has been harvested for decades and, and now with increasing, uh, usage of AI capabilities on the rise, artificial intelligence tends to be the trend. With recent, like product, uh, public releases of GP like chat, G P T AI is just gonna be more integrated to everything that we use.
Steve Hamm: Yeah. Yeah. So do you think ch g b T is, is gonna be used by data scientists and stuff like that, or, or are you just kind of talking a more general social and and economy kind of vision?
Mike Siswanto: Uh, I think it's more, uh, general, right? So like I, I think, you know, there's a lot of things that, uh, chat GBTs is really good at, right? So you, you throw it, uh, you know, a a, a large, uh, document to [00:21:00] summarize it will summarize it pretty well, right? Uh, you ask it. Really bespoke questions. Like for example, uh, you know, my daughter was, is a really big, uh, K-pop fan, so she was asking all chat, g p t, all these, uh, you know, bespoke, uh, uh, k-pop, uh, idols, uh, questions and it didn't get it right , right.
So it, it, it's, it's all about like, you know, being to improve over time with.
Steve Hamm: Yeah. Yeah, I'm sure it will, but it's kind of encouraging that it doesn't get everything right yet, you know? You know, um, hey, let's look even further out and, and I'm asking specifically about, you know, data management, data analytics, technology in, in business and stuff like that. Put on your visionary cap for a minute.
Look out five years or more. How do you expect these technologies to change business, the economy, or even.
Mike Siswanto: Yeah, very good question, [00:22:00] Steve. So if we continue a little bit about the talk of ai, right? In order to train these complex, sophisticated models, it requires very large data sets, right? And a lot of these data sets, they're private. and a lot of them have, uh, a barrier to entry and a lot of frictions to acquire, right?
So I'm hoping that, you know, in five years for a better frictionless data exchange that is agnostic to wherever the data is stored in, in whatever public, uh, cloud provider, and this will lead to, you know, essentially, , uh, better models for, uh, accurate results and able to cite how and, and where, uh, the AI formulated that answer.
Right? So, cuz I, if you know what it's like right now, they don't actually cite where they got
Steve Hamm: It's, it's, it's a black box. You don't, you don't have any idea. Yeah. But let me ask you this. If, if, you know, obviously a lot of these foundation models, they're, they're based on everything that's on the internet. That's everything [00:23:00] that's public. You, you mentioned a vast amount of information is privately held.
Uh, so. , it seems like it's one thing to kind of bridge between the two, but a second thing is how do you, I mean, there's a technology feat to be accomplished, but those, there's also a decision about data sharing and under what circumstances and things like that. It seems like there's, there are gonna be incredible governance, headaches or complexities.
If you really try to share, you really try to use models that go between private and. Information stores. So how do you see that worked
Mike Siswanto: Yeah. Yeah, that's correct. It, it's, it's gonna be a very delicate subject, right? Like, you can think about like, you know, uh, health data, right? Like, versus like, you know, uh, being able to tie an actual, uh, individual down, right? So like, like there's a lot of, um, PII and, and you [00:24:00] know, uh, hipaa. Constraints that, that you obviously need to abide, oblige by.
And so it's gonna be a very delicate and difficult task, right? So, um, it, you have to find that balance in that medium. And, you know, I don't have that answer, but it, it's gonna be a tough one to
Steve Hamm: yeah. People are really struggling
Mike Siswanto: these models Exactly. Cuz the, the models need data and, and now it's just more. How much, what type of data is it actually need, right? Like to, and, and what is it that we wanna focus the AI on and it, and, and, and make it, uh, you know, focused, uh, on, on that subject area. Does that make
Steve Hamm: yeah. No, that makes total sense. You know, you mentioned health. Anonymization of health data, kind of longitudinal studies and, and, and data sets. That's already something that's being managed, but it's, it's when you really start to look into data that might be proprietary to an [00:25:00] organization, but that a whole industry, everybody in the industry can benefit from it.
I think that's gonna be a great, a great opportunity, but also a, a pretty big challenge.
Mike Siswanto: Yeah,
Steve Hamm: be interesting to see how that goes. Yeah. You know, Mike, we're coming to the end. We typically end on a lighter note, a more personal note, and I understand that you are a golf nut. And that you also, you know you like, you like your technology and you're using IOT and data analytics gizmos to help you improve your golf game in real time.
So I think there are a lot of people listen to the podcast right now who are just kind of at the edge of their seats. They want to know how to improve, so give us the dope on.
Mike Siswanto: Sure, no problem, Steve. So, I mean, I didn't pick up golf until later on in my adult years. Right. So, and as you know, when you first start off, You're
Steve Hamm: you stink. Yeah,[00:26:00]
Mike Siswanto: so you, yeah, so, so you strive, you strive to be, uh, better by, by playing more, first of all, and then also taking some lessons, right? And so I bought these, uh, i o t sensors that you stick at the end of your clubs, uh, from driver all the way to putter called arcos back in 2019, right?
So at first, , all it did was collect data each round of each club you hit, and it knows how far you hit each club. And in addition, it kept your score. Right. So then, uh, they started introducing more features and, and, and they started applying, uh, AI on top of the datas that it started harvesting of for you.
Right. And they released this new feature called cad. , right? So basically this whole entire app, it knows your tendencies of each club. It knows how far you hit it, it knows, it tends to know how, which direction you spray. And, and it, and it, and you know it, it's able to detect where you are at relative to the green.
So it [00:27:00] also integrated weather conditions. Uh, so that includes wind now so that it's able to take into. What club to use based off of your tendencies and whether the the, the win is helping or hurting. Right? So, so like that's, that's like a very cool feature. When I first saw it, I'm like, wow, this is, uh, this is pretty neat.
This is pretty cool. And then it started improving, right? It even started improving because they started. You know, adding more features like strokes gain, right? So you're able to compare, uh, you're at a certain index and you're able to compare, hey, what part, uh, do I need to work on? Is it my putting, is it my short game?
Is it my, uh, irons? Is it my drives? Right? So you're able to compare. For your index relative to what skillset you need to improve, and then you could also compare yourself against other indexes to see what does it take to get there. Right? So it, I, I thought like it, that was pretty cool, [00:28:00] being able to apply the data that collects and, and, and putting AI intelligence on top of it.
Steve Hamm: right. No, that's absolutely fascinating. Yeah, this has been an interesting conversation today. I see parallels between that and just a moment ago we were talking about kind of real time, uh, insights for, you know, big money managers and stuff like that. And there, there certainly is a parallel between golfing and, uh, and back office, uh, pension fund management.
I didn't know that until now. So that's, that's kind of something new and fun. You know, I, um, it's been really fun talking to you, you know? One tends to think this is no insult intended, that that very old, very large banks are kind of like stuck in the technology mud. But it's really clear from, from talking to you that Northern Trust is really using technology in very aggressive ways, uh, to service clients and, and looks like tremendous, you know, insights and tremendous [00:29:00] efficiencies to the, to the customers.
So I applaud you guys for that. So,
Mike Siswanto: That's,
Steve Hamm: Yeah. Yeah. So thank you so much for talking to us today.
Mike Siswanto: Thank you Steve. Appreciate everything. Thanks for having me.