Kelly Kohlleffel (00:00)

Hi folks, welcome to the Fivetran Data Podcast. I'm Kelly Kohlleffel, your host. On this show, we bring you insights from top experts across the data community. We cover AI, machine learning, enterprise data, analytics, and much more. Today, we're thrilled to have Veronika Durgin join us. Veronika Durgin is a seasoned data leader with over two decades of experience in the industry. She currently serves as the Vice President of data at SACS, where she leads the company's data strategy.

She focuses on enterprise data governance and digital transformation to enhance customer value. Before joining Saks, Veronika held roles at Sonos, Indigo, and Boulevard. Her background encompasses data engineering, analytics, and data leadership. Veronika, welcome into the show. It's great to have you.

Veronika Durgin (00:45)

Hi Kelly, thank you for having me. What a wonderful introduction.

Kelly Kohlleffel (00:49)

Well, you have a tremendous background, and I'd like to just start by having you talk about some of the insights. You've been across multiple companies, you've helped them modernize, you've seen things, you know, hey, how do I get from here to there with my data stack, with my process, with my people, with my approach, all with an eye towards world-class customer experiences. Just give us an overview of that data background that you have.

Veronika Durgin (01:15)

Yeah. Um, I say I fell into data, uh, but I've been corrected because I spent my entire career in data. So I kind of haven't really fallen into data, but I started back, and I joke about it in 1999. So I got my first adult job as a junior DBA in 1999, kind of like why, because I watched, uh, programmers fix Y2K issues before, you know, the world was supposed to end that I didn't really quite appreciate what it was at the time. Just thought it was fascinating and funny. But yeah, I started my career as a DBA. I spent a fair amount number of years way over half of my career, just performance tuning operational databases. It taught me honestly, a lot of appreciation for well-designed systems, for thoughtful, you know, data modeling, for what it takes to get everything to work well and how to build systems in a way that are easy to support. 

And then at some point I transferred more into analytics, but everything that I've learned, keep saying that being a DBA was honestly the best bootcamp I could ever have. So now more in analytics space, I actually kind of responsible for the entire data lifecycle. Now we have, your operational systems, have analytical systems, you have your machine learning, AI, GenAI, you name it. We have it all, but yeah, it's been, it's been a fun journey. I love it. I love learning, loving my, you know, my job.

Kelly Kohlleffel (02:53)

Let me talk a little bit or ask you a little bit about, you look at the different organizations that you've worked with, and I'm thinking about, you know, maybe a luxury retail business. What are some of the key challenges or obstacles when you're trying, when you've got a huge variety of sorts, sometimes I call it source chaos or data source chaos out there. What are the challenges that you have in the organizations you've been with when it comes to data integration and ultimately getting to that final analysis or final data outcome?

Veronika Durgin (03:26)

Yes, that's a great question. So depending on the work and, for me, it was always like, I love data as an industry. And that's why you kind of like the, the breadth of data is what really excites me and the different, you know, issues and obstacles that different industry run into. And the, specifically for retail, you know, there's a different kind of like depth vertical, right? Issues. And you kind of hit the nail on that one. The issue mostly for most retails, not for all, except for like Amazons and Walmarts is not the like velocity of volume, it's the variety of data. Because retail e-coms companies have such a breadth of like, there's supply chain, there's inbound, there's outbound, there's websites, there's customers, customers are interacting in different ways via different tools.

So the, the big, you know, the one of the V's of big data, I know we don't talk about it, but it still exists as the variety of data and integrating that data is a challenge. It's a challenge, you know, kind of like, you know, I worked for software companies that built everything internally. It's been a challenge to make sure that every application, every microservice can connect somehow with data.

Kelly Kohlleffel (04: 46)

Yeah, I agree. I think too, and you know, in the retail space, and I'm certainly not an expert there, but I feel like that as close as you can get to having the decision-making solutions tools that you're using, provide real-time data so that I can maybe it's make an offer, provide, you know, predictions around what I might want to provide to an end customer or something like that. It seems to me it'd be really important. What's your perspective on real-time, near real-time data for making the right decisions and giving me as a consumer the right data at the right time.

Veronika Durgin (05:23)

Yeah, that's, I'll give you like the obnoxious answer of it depends. So as a customer and specific situations, and I'm talking as a customer, I want for whoever's on the other side to know me. I just get frustrated. I get impatient. You know, if I already bought something, I don't want this thing recommended to me again. So this is where, you know, real-time data, real-time solutioning is important.

Kelly Kohlleffel (05:51)

I imagine as you're talking to business owners to or business unit leaders, you know, we would all want it. You know, I give me real time. I need it. I need it. So I'm sure it's one of those things probably right at the top of list. You're drilling down into right out of the gate saying, okay, let's talk about this. Why is that necessary? And then trying to dial back from there because it can get very expensive in certain cases for certain types of applications.

Veronika Durgin (06:14)

We joke about it is like, I need a real-time dashboard that I look at once a month. it's, you know, like you, just like have a different conversation with your business users about what they actually really truly need and what problem they're trying to solve before kind of deciding where you need to be as far as kind of like data freshness SLA.

Kelly Kohlleffel (06:37)

Yeah, agree. if I'm a business user to I always you know I'm thinking about I need this yesterday, right? But what about the the idea of hey I can I can build something I can buy something I could build something exactly fit for purposes like I want I could buy. How do you how do you think about that yourself, and then how do you talk to business leaders and key stakeholders you're working with when they're saying, Veronika, I have to have this now?

Veronika Durgin (07:07)

Yeah, that's, I bring build versus buy probably like at least once a week up. It's, it's a constant conversation. There are a few dimensions around that, so when kind of like making, you know, build versus buy decision, we talk about obviously total cost of ownership, right? And there are all this like hidden things of like, you know, if you're building, you need to have a team, you need to have tools to build it and you'd have a team to support it. You need to have knowledge.

When you buy, have integration costs, have, know, scalability of pricing models, et cetera. But I think now what we really have to also consider is time to value. Meaning we, especially in like the past two years of kind of how GenAI pushed everything to move just a lot faster. There's this constant hype and pressure. We have to think of the problem we're trying to solve. We have to solve it now. 

We might not have time to build it, even though building it might be the right thing to do. So we have to think of like, and I call it, you know, the third B bridge, bring something in to solve the problem and get yourself time to build it in a different way, or perhaps to buy something else. So that's kind of like another thing that I, that I'm kind of like starting to push teams to think about as well.

We might have to buy temporarily, or we might have to build temporarily. So whatever the decision is, again, we want to unlock time to value, but we also want to make sure that it's in a way where it's easily replaceable. That's the other thing. Like our platforms become kind of like just a pile of Legos where we have to make sure that they're replaceable and swappable.

Kelly Kohlleffel (08:57)

Yeah, I love that. I feel like there are a lot of platforms out there that you can get the foundational pieces down and they can give you some of that extensibility that you're talking about that optionality that you may need to build the specific things, especially the vertical-specific things that you may need. But I don't have to build the entire thing. I think about like I would never build a, you know, a Snowflake or a Databricks today from scratch. Why would I do that? It's available today. There's no reason to, but I may need to build something vertical specific for fashion retail that adds additional value for my business on top of that. That's what I'm hearing you say to a degree as you're going through this. Is that accurate?

Veronika Durgin (09:43)

Yeah, actually I'm glad you brought it up. There's like so many, like I have a bullet-pointed list of things of why, why we would want to build. To me, I think engineering time is the most valuable. So we want to focus engineering time on capabilities that are unique or unsolvable by a tool. And unique, I mean, if it helps company have new value stream, if it helps business users, if it helps improve customer experience. 

These are the things that engineering time should be spent on, which means that everything else should be outsourced. It's like you outsource clean the house, maybe not the best use of your time, right? You should spend time on something else a little bit more valuable. So that's kind of how I think about things. We love to build. I love to build. It's so hard to step away from it, but sometimes you just have to outsource specific activities that are not necessarily valuable to your team, to your company. 

And I also, the other thing that I often say as well is, and think about something that's valuable for one company isn't for the other. We generally kind of also tend to lean like, I always wanted to build it at company ABC and now I'm at company XYZ, I'm going to build it. Well, it's no longer applicable. So it's kind of coming from, you know, what problem we're trying to solve. Is it unique to the business, should we solve it ourselves? Is it improving whatever that may be? Otherwise outsource.

Like we can talk about Fivetran because we're on Fivetran podcast. I keep saying copy, moving data from place A to place B is a solved problem. As a matter of fact, it is boring. It's not to me as a, as a like data engineer, you know, copy pasting more Python code. It's not exciting.

It's not. What's exciting is actually using all this data, putting it together and figuring things out. That is exciting, but maybe I'm like special, but I keep saying, you guys solved a boring problem. So.

Kelly Kohlleffel (11:55)

So no, that's good. I mean, there's, there's so many applications, applications of this across the board. I want to go back to Veronika, there's something you said, because I really picked up on this. You, said uniquely applying engineering time. I think of, you know, one thing that's, that's unique to every business is your data that nobody else has your data. Right. but I like you said, the way you apply engineering time can be unique and may change business. I got this very unique data, but I may not differentiate myself unless I apply my engineering time uniquely. Do you want to dig into that just a little bit more?

Veronika Durgin (12:32)

Um, I think like if I had to summarize it and generalize it, it's solving a meaningful, important business problem that is specific to the company you're working for. So perhaps, and, and I'm not sure Kelly, if this was kind of, you wanted me to take it. Um, let's speak on Amazon randomly for no reason whatsoever. Amazon's business problems are different. Different scale, different type of business, different customers from, I don't know, let's say Marshall's TJX, totally randomly picking, no affiliation with either. Different problems, different customers, different things to solve. So if I was an engineer working for Amazon and then I end up getting a job for TJX, how I approach things in how I solve and what I solve will likely be different. So I kind of always, like, you have to be in the moment. Your experience with fundamentals is important and critical, but the actual solution in what you're solving for might be completely different.

Kelly Kohlleffel (13:44)

Love that. How have you run into this? I know I'm spending a lot of time. It's just very interesting what you're talking about. How have you dealt with engineers or maybe architects that kind of want to build the cathedral, if you will, like, let's handle every single use. We got to do that. Got to, know, it's going to be a while, Veronika. It's going to take a while to get there. And how do you balance that? Because the business doesn't necessarily want this multi-month, multi-year type thing, they need something now. How do you, just from a team perspective, how do you keep those people engaged where they don't get discouraged and, but you don't go down that cathedral path, if you will.

Veronika Durgin (14:29)

Yeah, that's a hard one because I'm also the one who always wants to build perfection. So it's always like, whereas over engineering, I think, you know, like to me, agility is a mindset. It's, time to value now is so important. So what can we solve in two weeks, in four weeks? How does this fit into the bigger picture? We want architects to think two steps ahead.

Right. That's how systems that we built are supportable and scalable and resilient and interoperable, but we need to deliver continuously and immediately. So it's a balance. I actually, love having conversations with architects with like their heads on the cloud. I'm like, that is your job. My job is to bring you back to earth. So how do we find like, and I think we find that balance. I think to me, it's, it's important to have teammates on both spectrums of people who are kind of like just, I don't know, day to day, whatever, I'm just only immediate problem. People who are thinking five steps ahead because that's the way we find that happy medium and we move. And I oftentimes again say, what can we deliver now? And again, doesn't mean it's a lot of tech that quick solution. It's what we can deliver now. What can we deliver two, two weeks from now?

Are we generating a lot of tech that right? There's all these like dimensions and it's again, I'm kind of like back to it depends, Kelly. It depends. It's depends.

Kelly Kohlleffel (16:00)

I think you're right. There's there is this balance, right? I don't want to give you something in a day that, you know, breaks every time you use it. But at the same time, I have to you talk about time to value. It's so, so important. It's important when we're it's important from a tooling, people, and process perspective. Those have to work together.

Veronika Durgin (16:19)

And also, again, I'm going to bring now, like, you know, over, you know, the past 25 years that I've been in this industry, things were changing always. But when I started, we were planning to have the same system for the next three to five years. Now we have to plan to have the same system for the next two to three months. If we spent too much time planning and thinking, we'll be left behind by like so much that we will never be able to catch up. 

And it's like, really like it, it almost like we have to be on our toes all the time. And again, you, you said it too, but we can't rush either. It's like, it's a lot of kind of cognitive pressure to like find that balance. But I think I'm going to stress again, it's important to have people on the team that have these like opposing views. Almost. think that's the only way for us to truly like be pragmatic, but also keep up.

Kelly Kohlleffel (17:17)

Love that, love that. What about, I think we've kind of hit in and around AI a couple of times today already, but talk to me a little bit about what do you see as things we can do now, practical impactful applications of AI, especially in the retail space, luxury retail, anything you would like to cover there. I'd love you to, the now, what can I do now?

Veronika Durgin (17:39)

The now, great question. gosh, AI has been, sometimes it felt like drinking from fire hose. The hype is unreal. It's like, I don't think I've ever seen anything hyped so much. And it was hard to sort through the noise to get to real practical things.

But I think of AI broadly as kind of like three things, machine learning, which we've been doing for years and we're continuously doing, right? This is your predictive analytics. Continues happening across the board. Takes a little bit of time. You have to train the models. You have to have good data. You have to retrain your models. 

Then we have the new world of GenAI, which is again, incredible. It's your LLMs, unstructured data can do a lot of things with that. What I think you're describing is tools that use GenAI. So you want to outsource some of what you have to do to a tool, which to me, this isn't any different from any other tool. It's just now tools have that GenAI capability, which is incredible. But to me, this is just a productivity tool. So, and if anyone out there is not using a productivity tool at this point in time, I respect your principles, but I don't understand why. Specifically now internally for companies, again, like machine learning has been around for a while. It's been improved, fine tuned. We’re becoming more efficient. We're building out kind of like frameworks, processes around it. Tooling right. ML ops now is kind of a thing.

We're still stumbling to make it better, but this is like time to value specifically for ML, for GenAI, various use cases, real use cases, specifically, I think for any company that has customer service summarizing transcripts, like this is like a classic use case. And it's funny because it's only been like around for two years. I already call it like a classic use case. If you're not doing it, you should be, you know, every.

Kelly Kohlleffel (19:43)

Maybe, yeah, maybe. Yeah.

Veronika Durgin (19:44)

Every large, probably small to vendor brought in ability to use LLMs within their platforms. So it gives you kind of like it removes the need to manage infrastructure. So it lowered the barrier of entry. I hope everyone's experimenting. You know, there are options to, images is actually vision is another one that I'm very curious about. Like we can do a lot of analytics with them for images.

Documents is another classic use case, you know, PDFs, scanning, extracting data. So plenty of real use cases, but I think we need it like a year to sort through the hype.

Kelly Kohlleffel (20:26)

So interesting use cases in the retail space, either that you've seen or that you've heard about, I don't know, don't know, riff for a minute on, on, GenAI and LLMs in general, when you and I talked before, there was, there always is every day, there's just so much going on right now. You know, we see another time, agentic AI, right? Like how can I build this workflow or orchestration of agents to not just do a summarization, but maybe string together multiple things to help automate an entire process. Still too early in that space right now, I mean still evolving? Give me some thoughts on agentic AI.

Veronika Durgin (20:58)

I think we're a little bit far away. I think honestly we call some, I don't know, chat bots agents. I think the, the dream is good where, you know, these little programs will kind of talk to each other and figure things out. I am not quite there yet, honestly, like I haven't met, I haven't made a mental connection of how this will come together.

I think I've always been, of an opinion of data should find you when it's important for you to know about it. So something someone said to me once that data in your face, like that's the ultimate goal. Right. So this is where I think those agents will live to me. This is what they'll deliver. I don't want to have to log in into a dashboard. I want for the dashboard to find me if there's something for me to care about. Right. Like I think that's the dream.

But to riff a little bit, to be honest with you, I've, um, attended a couple of events. I am fascinated by what's being done in healthcare. Uh, there is like ambient conversation recordings when, you know, doctor patients, like right now you go to a doctor and the doctor just sit there for 10 minutes typing. Right. So there's like these machines that are recording ambient conversations and then they, you know, transcribe, summarize, so it's no longer dictating notes, it's no longer typing them. I think that was fascinating. I heard someone experimenting with like, I think glasses in operating rooms where they like look around and record all the like everything that's there, everything that's missing.

So like these images are then analyzed. I think it's incredibly fascinating. And maybe because I spend my day today with retail. I'm actually absolutely fascinated and inspired by what other industries are doing. I think it's just incredibly, incredibly, just interesting.

Kelly Kohlleffel (23:02)

I agree and you know what we do in this data space, I mean it translates industry to industry. If you've got an industry expert within an organization, I think you could take data teams and you know kind of plant them there and they would, the same concepts, the same things that you should do to make things as good as possible within a data program and an organization, it translates over. It gives us a lot of flexibility in this industry, you know.

Veronika Durgin (23:30)

But it's also like, if a problem's already solved in another industry, we shouldn't need to solve it again. We should just like take it and then build and improve on top of that. I've always been kind of like, my mind is leaning like philosophy, the way, learn from others. So you don't have to do it again and do something else.

Kelly Kohlleffel (23:50)

100%. I am, I am all about repurpose and reuse. It's something I talked to my team about all the time. All right. Let me ask you, Veronica, I know we're, running up against, I feel like I just talked to you all day on this topic, but, anything that stands out in, retail, you say, okay, kind of looking out over this long 12 month, right? You know, over the next year. Anything that you could recommend for, for data leaders to say, Hey, you need to be thinking about this, looking at this, considering this to make sure that we're positioned to, you know, stay in the, in the right place, keep ourselves ahead.

Veronika Durgin (24:24)

Yeah. And I know like you keep asking retail to me, data is an industry in general. I…just through my experience, I never, I would say of just outside interesting different data sets always kind of found that foundation of data and you know, methodologies and, people and processes is pretty like there's a big overlap. So I think in general to data leaders, to data teams, I would give maybe two recommendations. One is make sure you're working on things that help solve big problems. Like you absolutely have to align with company priorities. I don't believe for a moment that there is a data person out there who's not busy. There's a million of little things that always land on our plate. It's very, very important that we don't just stay busy, that we stay busy on solving critical and important high-value problems. That's how we stay relevant. Specifically, again, in general, AI, if teams are not experimenting and getting comfortable, they, they will just, it takes a while to understand how it works, what it does, what it is good at, what it's not. It's a bit of a mind twister where you kind of have to start thinking a different way that always takes a little bit of time. So even if there's no use case, even if AI doesn't offer you whatever, even if it's still expensive, find a little bit of time to experiment and just get comfortable because very soon it'll be cheaper, it'll be faster, it'll be better, it'll be whatever it is and then you'll have to be able to pick it up very quickly. So do your homework now.

Kelly Kohlleffel (26:05)

I love that. I think that, you know, if you're experimenting with prompt engineering right now, a couple of things I was gonna mention on this. Certainly, it will benefit you as you go along. Maybe you're building an application. Most of these applications you're going to have to build in the background, the prompt, right? And I was gonna get your thoughts on this. I find myself, Veronika, building the perfect prompt to build the perfect, like...Help me build the perfect prompt to build the perfect prompt. Is this, am I just crazy by doing this? But I'm doing, this is happening more and more and more for me now.

Veronika Durgin (26:40)

So I absolutely, but Kelly, think even still it doesn't necessarily like, we are communicating with a machine, and I think we have to learn to communicate with machine to be precise, to provide context. The other thing we joke about, I'm like, do you talk to your wife right now like that and your kids? Kind of like, yeah.

Kelly Kohlleffel (27:02)

Are you nicer to the machine than you are your wife or your kids? Yeah. Yeah. Yeah.

Veronika Durgin (27:04)

Or vice versa, you're very precise and verbose and detailed and your family is like, what is wrong with you?

Kelly Kohlleffel (27:12)

Yeah, we definitely shouldn't be nicer to the AI than we are to our family, right? Oh my gosh. That's a good, I love that. 

Veronika Durgin (27:16)

I hope so.

Kelly Kohlleffel (27:20)

That’s great. I love that. Let me ask one more thing. I got to ask this. So when you think about Veronika for you, most valuable lessons is across the board throughout your career, particularly as a female leader in the data space, anything that you'd like to call out that other enterprise leaders, male or female, can apply today to their role or what they're looking at within their current organization.

Veronika Durgin (27:43)

Yeah, I think for me, just growing up in the tech industry back when I started, I think resilience was probably the biggest thing. And I kind of had to be that anyway, by growing up where I did, by, you you know, the country, the, you know, the world, the time. And I think part of the resilience was just having thick skin. I don't get sensitive.

I just, it's, it's fine. If I'm in a situation I don't want to be in, I do my best to remove myself from that situation. I've always taken an approach. If I'm not welcome at this table, there is a seat with my name on it at another one. And I just, I don't, I'm not a fighter. I'm, you know, very kind of like determined, but I also don't stick around if I don't need to be there. So that was kind of like one thing that just helped me. I worked with some amazing people, men or women. I worked with some not-so-amazing men and women. I think as humans, we have to figure out a way to work together. And they just like, that's what I had to do, basically.

Kelly Kohlleffel (29:00)

Do you feel like you're born with that quality or that dimension of resilience or is it something that there anything we can do? Like if I said, Hey, Veronika, I'd like to really nurture and foster resilience in myself. I want to get better in this area. Is there anything that you can think of, or is just something innately that you're born with?

Veronika Durgin (29:17)

I think for me, like the environment I was in, I had to become that. And maybe it come a little bit more naturally for me, just of my personality. But I think in general, what I'm trying to encourage on my team is a level of psychological safety. And it comes with the fact that nobody is ever attacking anybody personally.

If it gets to feel that way, then the conversation should be had. Like, we should have conversation by it right away. Then it's just a conversation. There's no drama, and kind of, and again, I know there are people that are mean and they are, you know, attacking you personally. And hopefully you're, you're never doing that, but within teams kind of fostering that culture of like, we have to be comfortable with each other and knowing that all of us are trying to do the best.

Sometimes we have bad days. Sometimes we're very passionate about something, but it never comes from a place where I am telling you that you're stupid or whatever. It's never that. We're attacking problems. We want to come up with the best solution. But I know, you know, different personalities.

Kelly Kohlleffel (30:33)

Great wisdom, great counsel. I mean, you really never know on an almost a minute-by-minute basis what may be going on in somebody's life that may have caused them to just be off for that moment. Yeah, I love that. Veronika, this has been a lot of fun. I've really, really enjoyed having you on the show. I enjoyed talking to you when we chatted earlier and this has been great. I feel like we could do like a whole Joe Rogan three-hour thing here, but Fivetran won't let me do that. Thank you so much for joining the show.

Veronika Durgin (31:02)

Thank you so much, Kelly.

Kelly Kohlleffel (31:03)

It was a lot of fun. Looking forward to keeping up with everything that you're doing for the data community in general and at Saks. 

Thanks everybody for listening in today. We really appreciate each one of you. We'd encourage you to subscribe to our podcast on any of the major platforms, Spotify, Apple, Google. You can find us on YouTube as well and also visit us at fivetran.com/podcast. You can also send us any feedback or comments. We'd love to hear from you at podcast@fivetran.com. See you soon. Take care.