[00:00:03] Speaker A: Welcome to the True Data Ops podcast, everyone. We'll get started in a few seconds more to allow folks to get logged on to the live stream. See you in a few.
All right, welcome everyone to this episode and season three of our show, True Data Ops. I'm your host, Kent Graziano, the Data Warrior. And each episode we try to bring you a podcast discussing the world of Data Ops with people that are making Data Ops what it is today. Be sure to look up and subscribe to the DataOps Live YouTube channel because that's where you're going to find all the recordings from our past episodes. So if you missed any of our prior episodes or you want to re watch them, that's the way to do it. Better yet, if you just go to truedataops.org you can subscribe to the podcast and then you'll get proactive notifications about what we're doing, who my guests are, and what's coming up next. So today my guest is the VP of Data Cloud Architecture at Snowflake, Kevin Baer. Kevin was one of the first people I worked with and even did presentations with when I joined Snowflake back in 2015.
And he's still there pushing the limits of what we can do with the cloud and AI and all sorts of things. So looking forward to the discussion today. Welcome to the show, Kevin.
[00:01:29] Speaker B: Hey Kent, how's it going?
[00:01:32] Speaker A: Pretty good for folks that don't know you. Why don't you give us a little background on you and your work in data architecture and your career there at Snowflake?
[00:01:44] Speaker B: Sure.
I'll divide it up into three pieces because that seems to be how things worked.
I spent my first 10 years of my career really working with federal customers in the FSI space, working with different systems integrators, doing a lot of application and database integration work, basically connecting applications into databases. Worked a lot with Oracle Sybase. That was my first introduction to databases.
Really loved that. Right. I also figured out that there were people that did application development better than I did. So I eventually had the opportunity to do sales engineering to go start working for IBM. So I spent the next 15 years of my career at IBM really in the sales engineering and starting to get into software architecture, data architecture. I led software architecture team really down in the Southeast working with our largest customers around SOA. The Service Oriented Architecture had hands on with DB2 EEE. Right. We were doing the kind of the, the MPP space wars back in the, back in the day and then kind of the next part of my career. I was there for IBM for about 15 years. And the next part of my career was really with Snowflake. I was looking for something different. The opportunity to work with big data, work in the cloud. And what better opportunity than to join on with Snowflake? And I've been there, coming up on 10 years. It's been a really fun journey and working with large customers. I think the thing I like the most really about Snowflake, obviously, besides the fact being the biggest IPO in history, it was.
It's really when customers understand what Snowflake is and how it operates and why it's different. Kind of, you almost see the light go on in their eyes.
That is exciting, right? When you, when you're, you're part of that journey, you're explaining how it all works, you help them kind of get into production. That, that has been such an. The exciting part, at least in my journey at Snowflake.
[00:04:06] Speaker A: Yeah, I mean there was for me as well. I hadn't realized you started in the federal space. My career started at Department of Interior out in Denver doing building apps, even doing stuff in Fortran and basic. And then that's where I got introduced to Oracle and I built, got to build some of the first oracle applications for U.S. geological Survey and Bureau Land Management and all of them. And that got me into the databases and got know, got very interested and eventually moved into warehousing like, like you did and getting into the big data space and then on to Snowflake and yeah, those first couple years, especially working with Snowflake and going out and like you say, going out and seeing the light bulb go on as you're explaining how different Snowflake is and how it works and the possibilities that we now have that we didn't have in the traditional on prem architectures that you and I both worked with. Yeah, that was very exciting. So tell me a little bit about this role you have now.
[00:05:07] Speaker B: Yeah, so I recently moved into a data architecture role. So if you think about Snowflake as a platform, I think we've had a lot of success going to market, solving data problems in particular. Now I think we're starting to elevate, we're starting to get into conversations around data architecture. Right. And over the last, I'd say five to 10 years, you've started to see some different architectures come into play. Right. Whether it's a data lake or a data mesh or all of these different architectures. I think the thing I got excited about with Snowflake was talking with customers.
Snowflake is different. Snowflake actually gives you the ability to think about data architecture in a different way. So most recently, I published a paper basically putting together what we call the data cloud architecture, not necessarily specific to Snowflake. You can implement a data cloud architecture even without Snowflake, but it was really around some of the principles Snowflake really allowed you to do. The ability to move data or interact with data between, not only internally, but externally or with your partners, basically to interact in a completely different way.
My team and I have really been focused on how do you do that? How do you think about data differently, not only from an internal management perspective, but from an external management perspective. It's really becoming an interconnected world. And how do you do that? How do you manage that whole interaction?
[00:06:42] Speaker A: Yeah, yeah. And so right up there with having to understand the internals of Snowflake architecture and how that allowed us to do things differently now, from a bigger perspective, really, how to tie this all together with all these various architectures out there. Because one of the things that I enjoyed a lot, especially when the data mesh stuff first came out, is realizing, well, you've got options now, and Snowflake allows you to choose the architecture you want that's going to work best for your organization, whether it be data mesh or data vault or a data lake or a data lake house is. Our customers had all of those options, right?
[00:07:20] Speaker B: Yes.
[00:07:21] Speaker A: And having to think about it was definitely a challenge for some people, right?
[00:07:27] Speaker B: Exactly. And Snowflake allows you to implement whatever you want. But the other thing I started to see a lot of with customers is there isn't one data architecture that fits all. Each one of these architectures is trying to solve a different problem, and you can even smash some of these architectures together. So it's not like you have to have one architectural paradigm. You can do multiple architectures, even within a line of business or across lines of business. You know that that's perfectly fine. Right. You're trying to solve a specific problem with each one of these things and that. And Snowflake allows you to do all that.
[00:08:02] Speaker A: Yeah. And so that's probably a good segue into what I wanted to talk about today, because in this season, we're trying to take a step back and kind of think about how the world of what we call true data ops has really evolved with all of this, and what have we learned in the last few years? And you're definitely a good person. You know, 10 years now, basically, at Snowflake, like, a lot has changed, even since first days of Snowflake. So. And you've helped customers implement. You talk to customers all the time and trying to help them take cloud in general and of course, Snowflake in particular. So can you tell me a little bit about how you've seen this space evolve in the last few years and where you think that might be going? You know, what do we need to do and. And to make things even more successful?
[00:08:50] Speaker B: Yeah, I think we're still kind of in the middle of this transformation, right. I mean, in my lifetime, and I think your lifetime, right. We've had the Internet and we've had really public cloud infrastructure come into play. Right. And a lot of these companies have been around for a while. They've all been operating in data centers. And so now they're trying to figure out how do we leverage not only public cloud infrastructure just to reduce costs. Right. I think that was the initial thought process. We're just going to do all the same things in our data center and we're going to basically shift that into the public cloud and do the exact same thing. We've had the pleasure, Right. The ability to talk about, well, the Internet and public cloud gives you the ability to do things differently. Right. And I think right now you've gradually seen, you've seen some customers have gone whole hog, right? They've gone all in on the cloud right away. But most of our really large customers are still in this transformation. They're still in this transition to cloud infrastructure. And a lot of times it ends up being a hybrid infrastructure. You have to work with them, meet them where they're at. That's really, I think, the key piece. If you basically go in with a message, you should do everything in the cloud.
That's going to meet with instant resistance. For a lot of customers, that doesn't make sense because they've been working in data centers forever. So just kind of evolving the thought process not only from don't do the same thing on premise that you do in the cloud, you have so much more capability, you know, the elasticity of the cloud, being able to take advantage of it. But I think now what's happening is obviously the whole conversation around AI and gen AI, right? You can't do all of the things you want to do kind of within the data center without having all of that elasticity, all of that processing power, all of those GPUs, that all exists in the public cloud. Right. And so that conversation is always ongoing. So I'm seeing more and more people, more and more customers want to take advantage of that. And that means you need clean data. Right. You need the ability to process data from multiple sources, get it into a format where you can start to use it to do all of these new, unique and innovative things.
[00:11:15] Speaker A: Yeah. So with that in mind, so from your perspective, what you're seeing, what do you think, what is DataOps to you and how does that fit into helping customers manage this really complicated data and sometimes hybrid data landscape?
[00:11:33] Speaker B: Yeah, what's funny, it's not only hybrid, just from a, you know, whether you're on premise or in the cloud, it's hybrid tool, you know, tool sets, Right. You've got to integrate all of these different tool sets to get to an outcome. Right. And when I, when I think of Data Ops, the outcome is getting data or applications available for use. Right. By your end users. Right. So you have to take all of these different data sources, you have to figure out how to consolidate, you know, put it into the right format, get it all, get it all ready, and then get it into production. Right. Whatever production means to you. And that is a challenge. Right. And that, I think that's where the whole Data Ops framework, all of the things you guys are talking about makes sense. Right. And it's easy to do that, like, say, for one data engineering pipeline or for one thing, it's a lot harder to do that at scale. You know, 10, 20 different product integrations, hundreds of different pipelines that you're creating, and then your end users want to consume in all kinds of different ways. So when you put, put something into production, you've got to be thinking about it in all of these different dimensions. And I think DataOps allows you to do that.
[00:12:48] Speaker A: Yeah. And it's. We ran into this in the software world and you had the application development experience as well.
You know, DevOps tried to solve that over in the software world. And this is now, okay, how do we do this in the data world and make it coherent and make things really make sense and make it manageable? And you said a key word there at scale.
That's the key. Like you said, one data pipeline in one database with a couple of engineers, you might be able to keep track of everything and make changes and push bug fixes into production without breaking anything. But when you start getting all these different technologies involved, because like you said, there's depending on what you need, there are different tools. And then Snowflake obviously has a massive partner ecosystem because there isn't one tool does not solve all the problems for.
[00:13:47] Speaker B: Everybody and it never will.
[00:13:49] Speaker A: Right. Yeah. So you know, for, for listeners who, if you've not looked at what we're kind of talking about here, the True Data Ops, there's the seven pillars of True Data Ops. You can find
[email protected] 7-pillars or scan the QR code that's on screen right now.
Now it's been, it's really been four years. It just is mind boggling to me actually since we first published the, the truedataops.org site and did the Dummies Guide to Data Ops, I was still at snow when we, when we were kind of working on the first cut of this with our friends Justin and Guy who at the time were at Datalytics in the UK and they were the first SI for Snowflake in Europe.
And that's where this all started and it evolved.
So we've got these seven pillars. I'm wondering, you know, from your experience, do you think that those seven pillars are still resonating today with all of this gen AI and everything else that's going on?
[00:14:52] Speaker B: Absolutely. Right. I think even more. And when I'm kind of, the way I'm thinking about it, the way I'm seeing it, it's not just whatever you're trying to, you're always trying to get, again, get it from operational into analytic, get it in a way to analyze all of this data and do all of these things. It really doesn't matter if it's just data or it's analytics or what it is.
You need a way to think about all seven of those pillars, right? You're, if you're focused on just one, you're gonna, you're gonna miss something. Right? Things aren't gonna come together the way, the way you wanna do it. Right. And I do think they, they all resonate and they all work really well together. And implicitly, even if you, you know, even if you didn't know about the seven pillars, you kind of end up doing all those anyway, right. If you, if you didn't think about it that way. But it's nice to be able to, if you can break it into something that's, that's workable, right. And you can think about each one individually and then piece it all together, you're going to be much better off.
[00:15:53] Speaker A: Yeah, yeah, you said.
[00:15:54] Speaker B: Yeah.
[00:15:54] Speaker A: There's all of those parts there that we think about, whether it's the doing the security and governance, making things as we used to say, modular. We used to do modular programming, right. Breaking things down, environment management. I mean, you And I, I know we had lots and lots of discussions about how do we do dev, QA and prod in a Snowflake ecosystem.
[00:16:20] Speaker B: Absolutely. Yep.
[00:16:21] Speaker A: And you know, early on we had, you know, multiple architectures, right? Multiple recommend ways that customers could do it depending on whether they wanted to have one account or multiple accounts. And managing all of that was different. You know, in the, in a single account you could use zero copy cloning. Well, if you're going to go between accounts or even now, different providers, Snowflake on Google versus Snowflake on Azure or aws, there were different things you had to do. And over time Snowflake has evolved really good tools for helping that. But you have to think about it.
It's not a no brainer, I guess, right? Yeah, it's not completely a no brainer.
[00:17:02] Speaker B: It's not a new brainer and you can think about it differently. Kind of like what I said before, maybe you were doing it a specific way in your data center or on premise. Snowflake. And not just Snowflake. But all of these tools really do give you a way to start to think differently about how you get data from one place to another, how you manage metadata. I know we're probably going to talk about that later, but how you do all these things, you have to be willing to adapt to the new technology, to the new ways of thinking to take advantage of those things. Right? There, there really is an advantage to doing things differently. And the whole goal, and I talked to a lot of people about this, right, Is getting data more quickly, right. I rarely talk to a customer that says, yeah, you know, I want to get my data more slowly from, from my, from my operational data sources, you know, and it's like, you know, they, they always say, okay, if I get it within the next hour, that's great. And I go, okay, that's fine for now. You're going to be back to me in six months, a year, and you're going to say, I need it in a minute. I know that's what's going to happen, right? And almost nine times out of ten, right, it's like, yeah, we were doing it that way, but now we need it almost near real time. Right? And it's like we could have started.
[00:18:23] Speaker A: There, but that's planned for it, right?
[00:18:25] Speaker B: Yeah, you got to plan. That's fine, we'll work our way there. But yeah, nobody wants data slower, right?
[00:18:31] Speaker A: Right. Yeah, exactly.
Yeah. I remember early on, some of the bigger customers we were working with when we had these kinds of conversations about environment management and all of that was like, well, we need three separate Snowflake instances.
Why?
[00:18:47] Speaker B: Right, that's why.
[00:18:48] Speaker A: Right. We need dev Prod and qa and our security policies say they have to be separate. It's like, that's because you wanted them on separate servers, right? Yeah. This is the cloud, Right. That concept doesn't apply anymore. And I remember us going round and round with some customers on that now.
[00:19:07] Speaker B: Yeah.
[00:19:08] Speaker A: There's, there's benefits both ways.
[00:19:10] Speaker B: It's. It's still happening. It's still.
[00:19:11] Speaker A: It's still happening.
[00:19:13] Speaker B: Still, still happening. Everybody, everybody has to go at their own pace. Which, which makes perfect sense. Right. But we're still having a lot of not, not as many of the same conversations. Right? We've gotten past a lot of that because a lot of customers have already solved that problem. Right. And so if you are, if you know customers have already solved the problem and you're very comfortable with that, why do you have to solve the same problem again? Right, right.
[00:19:38] Speaker A: So with the advent of AI and gen AI and all of that that's happening and, you know, Snowflake's obviously getting very big into that. Do you think these pillars are more or less important than they were before?
[00:19:49] Speaker B: Oh, more, more, more, more, more and more. Right. Again, same thing at scale. Right. I mean, you can't. Unless you have clean data from all of your operational systems. You know, if you, if you operate, we all know if you operate on bad data, you're going to get bad results. Right. There's just no doubt about it. Right. And kind of the way I think AI is solving very interesting problems. Right. We started in the, with BI space. Right. That was fun, right? You know, asking different questions in different ways.
And now AI is another type of analytics, but it's all the same. If you have bad data, you are not going to be able to do the types of analytics and get the answers that you want. Right. So this whole idea of data ops, right. Making sure you're doing things in the right way, you're cleaning up the data all along the way, you're managing it in the right way in order to get the results you're looking for is absolutely critical.
[00:20:49] Speaker A: Yeah. I think one of the places that in the data world we've fallen short for literally decades is the automated testing, automated regression testing and monitoring. Because. And we're not just talking about the code now, we're talking about the data. Right. Getting unexpected results in your data, especially now we're talking about AI could really skew the output of the AI, Right?
[00:21:17] Speaker B: Absolutely. Yeah. No, and people, people are still trying and we're still early in the process. We're still trying to figure that right. For some things, those wrong answers, okay, maybe it leads you astray, but you can't completely rely on the data. It's still just an indicator.
In other cases, if you're trying to put a model in place for fraud detection or whatever, the quality of the data and the quality of the model matters a lot.
You got to think about what your use case is and what problem you're trying to solve.
[00:21:53] Speaker A: Yeah, yeah. I think that's why that, that particular pillar, I think, resonates with me a lot. And trying to, to get people to understand, it's like. Yeah, it's not just throw the data on. I mean, when we started in the early days of data lakes, just throw all the data in there and we'll do all the data science on that. It's like, well, how good is that data? Do you know?
[00:22:15] Speaker B: Right.
[00:22:15] Speaker A: And you attach to a new data set. It's like, what are your boundaries? What are your domain boundaries here? Are you getting good data? And how do you check that? And I think that's one of the reasons that pillar is there is over the years, I definitely saw a lot of challenges even just building data warehouses that you'd get the business, get the BI report, they'd run it and go, this is all wrong.
This data is not correct. And you'd trace it back and go, okay, well that's what we got. That's what came in. And there's like, well, that's wrong. All right. We should have known that when we staged the data, not all the way through the transformation and through a data vault and coming out a data mart on the other end. And then they say the data is wrong.
[00:22:58] Speaker B: And that's why.
[00:22:59] Speaker A: How did we miss that? We just were really, really bad at.
[00:23:02] Speaker B: That's why we have spreadsheets. Right.
Everybody wants to manage. This is my core data set. I know this data is okay, but if everybody does that across the organization, everybody's going to get different results. It's. It's just.
[00:23:17] Speaker A: Yeah. And I think with the scale now that the impact of these type of errors is magnified, right?
[00:23:28] Speaker B: Absolutely.
[00:23:29] Speaker A: Yeah. It's one of those, you know, if, if you're just a little bit off path and you're only going this far, it's not so bad. But if you're going way out here, I mean, now you're like three miles away from where you were heading.
[00:23:40] Speaker B: Yep.
[00:23:41] Speaker A: Because you just. Because you're just off just a little bit.
[00:23:44] Speaker B: Yeah. And I would add that the other trick there is democratization. Right. Yeah.
We're very much on a democratization thought process. Right. I think everybody's understanding the value of getting the right data to the right person. You know, not too much data in a secure fashion. Right. But you do want to get more data out to individuals in order to make good decisions. So the importance of how. Of that data, the quality of that data to now maybe even new user populations that didn't have access to that data before.
[00:24:16] Speaker A: Right. They might not even know. They won't know themselves whether it's good or bad.
[00:24:19] Speaker B: They don't know.
[00:24:20] Speaker A: Right, Exactly.
[00:24:21] Speaker B: They assume it's good. Right. Because it's coming from a trusted source.
Now that whole process of making sure the data is accurate is so much more important. Kind of like you said, because of the scale, because of democratization, because of how people are thinking about how they're going to analyze this data. And I always talk about, you want democratization. Right. If you have just a handful of data scientists that are looking at things and making decisions, they don't necessarily have all the information that the business has. Right. And all the more these. We can get the tools and the data into the hands of people that are making business decisions, they can think about it differently. Right. So democratization in my world is a good thing. We're just trying to do it from a snowflake perspective in a governed way. Right. In the right way. If you have a data lake. Sure. Could you give everybody access to all of the data in the late. In the data lake? You could. I wouldn't recommend it. Right. Because there's probably a lot of stuff in there that you don't want everybody to have access to. So you have to do it in a governed way. And, you know, fortunately, Snowflake does that really well.
[00:25:28] Speaker A: Yeah, yeah, yeah. The governance and security is paramount these days, especially when you're trying to do things like that. Because. Yeah. Not only is there data that everybody shouldn't have access to for variety of reasons, whether it's pii phi data or just some regulatory restriction, some of that data just could be flat bad.
[00:25:49] Speaker B: Right.
[00:25:49] Speaker A: And if it's not been curated. Yeah. You do really want to be exposing that to.
We'll call them citizen data scientists, citizen data analysts that are out there that were business analysts that are now using data. Right. That don't necessarily have the sophistication and understanding of, you know, the ins and outs of, of some of this data that they're just, like you said, relying on it. It's coming from their, their source of truth and they just assume it's right and, you know, could end up making some bad decisions. And AI, of course, same thing, feeding that into an AI without any curation or governance on it, you know, could, could lead to some, you know, very serious unintended consequences.
[00:26:32] Speaker B: Yeah. Everybody assumes, and I always say, don't just assume cause and effect just because the data is telling you something. Make sure it makes sense to you. Right. Don't just rely strictly on the data.
[00:26:44] Speaker A: Yeah. So as we're talking about that, you know, where do you see, you know, metadata and even, you know, data catalogs playing in on this in the AI world?
[00:26:55] Speaker B: Yeah, metadata. You know, I love talking about metadata. I think it's core to everything that we're talking about here. But from a, from a kind of catalogs perspective, what I'm seeing is kind of the proliferation of, you know, I have some customers that have, you know, 200,000 tables. Right. Just massive amounts of data in their infrastructure. Right. And, you know, petabytes, even exabytes of data. So this idea of a catalog basically taking that data set and creating, I would say, highly curated data sets, whether you call it a data product or a data asset, I'm starting to see that concept coming around where you take these core thing, maybe it's your customer data, maybe it's your product day, whatever it is, these core data sets that the whole organization can use, managing them in a special way. Right. And if you're going to do that, then you have to add more metadata. You have to add more context behind what this data is. So that's why metadata is so important. Data with data without metadata or without context, it's just noise. Right. You can't, you can't make any, any sense of what it is or why it's important. That's why metadata is so important.
[00:28:13] Speaker A: Yeah. And I think even one of the pillars in Data Mesh about data products is that it needs to be discoverable and understandable and all of those things. And that's really, I'll say that, you know, it's more business speak for what we call metadata. Right.
[00:28:28] Speaker B: It's got to be there. Without metadata, you can't do any of that. You can't see, you're not going to be able to find, even if you know what you're looking for. Right. You're not going to go into that petabyte of data or all of that Data and do a generic search. I mean, it's just hard.
[00:28:43] Speaker A: Yeah. And so if it gets curated and productized, then you can talk about things like marketplaces and making things available that way versus just a data catalog, which is, you know, I'll say the 21st century version of a data dictionary that you and I were always being asked to produce in the past.
[00:29:01] Speaker B: Right, the semantic layer. Right. Take.
[00:29:05] Speaker A: We need to do a whole other show on semantics.
[00:29:07] Speaker B: Yes, I know, but semantic layer, it's fun, is now having different context, right? Because now, especially in AI, you're trying to create, you know that you're trying to get the AI to understand the data. So it almost needs its own semantic layer, which is, which is in essence just metadata.
[00:29:25] Speaker A: Right. It's all metadata. Just, you know, we used to differentiate between business metadata and technical metadata and it's. Yeah, now, yeah, now you're talking about, you know, what kind of metadata does the AI need?
[00:29:36] Speaker B: Yeah, yeah, yeah.
[00:29:38] Speaker A: Is that, is that the same semantic layer that business person needs?
[00:29:43] Speaker B: I think that's exactly. That's where we're at right now. I think that's where what everybody is trying to think through. Right. Because I have, I know I have a catalog, but is that catalog, can I actually generate some of that metadata and use it in the same way that I'm using it for my gen AI tools? Are the, are they the same thing? And it's actually similar, but, but different. At least in my head right now. But should those worlds come together? They probably should be because again, talking about, you know, data and automation, right, you, you really do want to automate a lot of what's happening there, right? Because if, the more you have to write on your own or come up with your own own tags or do, do whatever it is you're, you, especially at scale, you're. It's going to fall down, right? It's, it's obviously going to fall down. So the more you can think about that ahead of time, the better.
[00:30:33] Speaker A: Yeah, well, unfortunately, as I expected, we are like at time already, so. Yeah, yeah. What's, what's, what's next for you? What's the best way for folks to get in contact with you?
[00:30:49] Speaker B: What's next for me? That's always a good question. I am, I am still on this journey, Right. I think data ar, you know, data architecture for me and for Snowflake, I, I think is key. Right. We're again trying to meet customers where they're at, help them understand how to, how to solve all their problems. I really I'm a problem solver. Right. I, I just, I, I love solving problems. So if anybody wants to get a hold of me,
[email protected] or on LinkedIn.
[00:31:18] Speaker A: We have a QR code for that one up there.
[00:31:20] Speaker B: Yeah, love solving, solving data problems and pushing the envelope. Right. I mean, AI is one thing right now. In the next 10 years, it's going to be something else. Right. So I just love solving data problems.
[00:31:31] Speaker A: Yes. You mentioned you're working on these data architectures, concepts and all that. Are you publishing that anywhere?
[00:31:39] Speaker B: I did. I published the data cloud architecture on Medium.
So if you want to go find it, it's out there and that's where I'm at.
[00:31:48] Speaker A: All right, awesome. Well, thank you so much for joining me today, Kevin and I appreciate the insights and the conversation always. And it's like you and I could talk, we could keep going literally for days about all these topics.
Thank everyone else that's online for joining today or those of you who are watching this on a replay.
Be sure to join me again in two weeks. My guest is going to be AI and machine learning expert. She's the CEO and founder of Illumix, AI Ina Sela. So she'll be joining me here in two weeks and that's going to be a really, she's got some really exciting ideas along the way of semantics and AI.
So that's going to be a really good show. So as always, make sure you like any of the replays here from today's show, tell your friends about the True Data Ops podcast. And of course, don't forget to go to truedataops.org subscribe so you don't miss any of the future episodes of this podcast. So until next time, this is Kent Graziano, the Data Warrior, signing off. See you soon.