Kelly Kohlleffel (00:00)

Hi folks, welcome to the Fivetran Data Podcast. I'm Kelly Coleffel, your host. Every other week we'll bring you insightful interviews with some of the brightest minds across the data community. We'll cover topics such as AI, machine learning, enterprise data and analytics, data culture, and a whole lot more. Today, I'm really pleased to be joined by Moin Haque. Moin is a seasoned leader in data analytics and technology. He's got extensive experience in driving data strategy and innovation. Currently, he stewards enterprise data analytics and AI at IFF. He's helping enable, align and empower business and functional teams to unlock opportunities using data and AI. Moin, it is fantastic to see you again. Welcome into the show.

Moin Haque (00:49)

Thanks so much, Kelly. Pleasure to be here. Really excited to have the conversation today.

Kelly Kohlleffel (00:52)

I am as well. Well, Moin, to kick things off, I'd be interested to hear more about your role at IFF and your background in the data space overall.

Moin Haque (01:03)

Sure. So at IFF, I sit within the technology function of the company, and my focus primarily is stewarding enterprise data analytics and, more recently in this past year, into the AI space as well. And for data, for us and not just IFF the past couple of years, but throughout my sort of background in the data space, it's often the most common refrain that I hear in all the conversations with business teams and function teams is probably the most, it's a polysemous word in any company's vernacular.

But I think what's happening in a lot of the conversations is that data like anything else, it's an ingredient in the larger recipe. And whether it's a recipe for we're trying to solve for growth, we're trying to solve for productivity or innovation, what I've learned is we really need to hone in on where it's coming from, where it's coming in, how much of it, how it interacts with other ingredients and really not lose sight of what we're trying to create, who we're trying to serve with this outcome, what are their respective palettes or preferences. I'm getting a little carried away with the cooking metaphor here. So if I just say, we're not going to get there without data, and we're learning it's data that's going to make sure that where we're trying to arrive to, we get there in one piece, or get there at all.

Kelly Kohlleffel (02:27)

Can you talk a little bit more, kind of dig into in addition to that data ingredient, what's really important to get things right in a data program across an organization?

Moin Haque (02:37)

Sure. Yeah, I mean, the other ingredients for me, it really comes down to foundationally the people and subsequently the processes. Often data is really the outcome of that. It's the exhaust of that. It's the processes that are in place. It's the interactions of the people. And not having a lens on that and just looking at the data, that's where oftentimes I think we have such a disconnect with it. The data is not right. It doesn't fit.

And, to me, the data is really the innocent party here. It's really understanding how the other things interplay. So for me, it's really stepping back and contextualizing the processes. How did we get here? And ultimately, what are we trying to answer? What are we trying to solve for? Are we trying to move the needle forward? Is it something as simple as getting an insight and really getting that holistic lens? For me, it's key because then we can really figure out how bad is the data, how disconnected is the data, what is it going to take to be good enough, where are the gaps and so forth.

Kelly Kohlleffel (03:39)

You and I have chatted in the past about just enterprises in general, this proliferation of not just data, but data sources and all the different angles that data is being generated and coming into the organization, and how do you deal with that to get that final ingredient that is going to serve the needs of the stakeholders that you have? And I mean, it just gets the larger you grow, the more challenging it gets. How are you thinking about that today in terms of your current data stack and ultimately that flow down to your internal teams and customers?

Moin Haque (04:16)

Sure. So I can't talk about specific technologies in that sort of data stack other than obviously our amazing data movement platform, which we can talk about. But we're not unlike many large organizations who've come through into the current shape through various transformations, acquisitions and so forth. So you have a very fractured landscape, a very federated landscape. And it's the usual footprints, multiple platforms, whether they're commercial, whether they're custom, that are smattered across functions and operations. And I think some added elements for us being a global manufacturer, having an R&D footprint is that now those platforms and those signals are coming not just from one place, but they could be coming from factories, they can be coming from labs, they can be coming from public clouds, private clouds. So you have a real sort of collection of disparate signals at different grains, at different cadences.

And so that typically becomes the landscape, what we would call a very fractured data state, if we were to look at it. And that's not counting all the external signals that are also coming in. So our stack ends up really being a collection of these technologies, and each of them is evolving and changing. So it's very hard to really pin it down and say, here's what that architecture flow, that diagram looks like. So it's really being able to navigate in something that's going to be very fluid.

And the good news there is that it allows us to then say, well, let's really focus on the signals themselves, what matters, where it's coming from, and how that's all going to tie in.

Kelly Kohlleffel (05:52)

Is this the biggest challenge that you see? A fractured or fragmented data environment or just environment in general? Or are there some other things that we should be thinking about as well?

Moin Haque (06:02)

I think on the data side, it's almost just accepting the fact that it will stay fractured and desperate. I think the aspiration is not bringing it all together in one place. I think for me, what I aspire to is federation and federation that's sustainable and manageable and scalable. To me, federation is really the aspiration. And you realize that things will continue to be fractured, they'll continue to be disconnected.

They'll continue to overlap, and they'll continue to change. Things aren't staying static, right? You're doing M&A activity, you're doing divestitures, you're changing your business models, your markets. So if there's a way that we can have a way to federate and capture those signals across those various platforms, I think that really helps posture us for being able to unlock value and manage this better. And that really leads to this idea of what's almost more important than the data itself is the metadata about the data.

Kelly Kohlleffel (07:04)

Over the last probably, I don't know, 5 to 10 years, we've been systematically, I think, thinking about the way data moves into an organization and ultimately presents these signals as you talked about moving from a ETL pattern to more of an ELT pattern as you have looked at the industry and worked through various both industries and organizations, what have you seen? What's working? How does that fit into your overall strategy?

Moin Haque (07:33)

It's a key point. I mean, an ELT pattern is something I've been seeing now, let's say, for many years. We've seen that shift, that transition, especially as platforms and capabilities have transcended away from the limitations of storage or compute or even to some point network. You get that elasticity across all three dimensions, an ELT pattern really starts to make sense. I think the one piece I would say that's a little bit more than just saying, let's extract, let's load and then transform, is how ephemeral can I make that load? Do I have to physically copy the data all into the same place and then transform it there? Or is it something where I'm aware of the data, I know how to extract it, I know where to load it from, and then transform it in the way I need it when I need it? And that's the other shift on ELT we're seeing. 

Just to give an example, there are platforms out there where you'll hear the term bring your own lake. So I'm not necessarily moving my data, but I'm making that data aware and digestible and interoperable to whatever that platform is. So I think that's sort of the other pivot on top of the ELT shift that we're seeing.

Kelly Kohlleffel (08:42)

Yeah, I agree with that. mean, there's what you're describing. I mean, there are a lot of benefits to that type of approach. I think about things like, you know, avoiding vendor lock-in. I think about things like letting me bring whatever I want to plug into that lake storage or where that data resides. I can plug in the compute that I need at the time that I need it for the use case or workload that I need it for.

Is this something that you feel like is over the next, you know, couple of years is really going to progress as the way to do things?

Moin Haque (09:17)

Yeah, and I think as platforms become more interoperable and we start to see some standardization around being able to share these signals and share them faster, I think that's going to go a long way. We're almost coming to a point where the more modern VHS beta battle between Delta and Iceberg is starting to subside. Right? So it's trending toward that. Just even as recent, I think, as this week, we saw the ability to maintain tables and platforms like Snowflake outside of Snowflake Go GA. So it's starting to go that way. It's becoming less about I have to physically move things. How do I interoperate with them? And again, it all keeps going back to that metadata. I need to know where that signal is, how it is, so that I can marshal it when needed. And it's not a fire drill at that time to go look for it and then build a pipe to get it.

Kelly Kohlleffel (10:08)

I'd love for you to dig in a little bit more on how you personally, how you think about AI, GenAI, ML and, and, know, how, how your organization today and other organizations you see them approaching these types of implementations because it is accelerating in a big way right now.

Moin Haque (10:28)

Yeah, so I'm sure ourselves, like many organizations, what we're seeing is that it comes down to a lot of the same fundamentals. It's the fidelity and accessibility of our organization's knowledge, whether it's data, whether it's other types of insights, that's really becoming, going back to the earlier concept, that's the key ingredient. That's the key role, the key differentiator in how we're going to get applications and opportunities iterated using these new capabilities. 

I think one of the things we're trying to avoid is not look at this as this is the tool that's going to solve all the problems. All of a sudden because we're going all in on AI, we're forgetting about all the other disciplines that we were doing around data, around insights, around process transformation, around, remember, all the digital strategies. So it's really kind of figuring out how does this fit in? How does this enable and going back to the basics, which you would do with any technology or capability. What's the value here for the organization? Is it, you know, an AI, we typically are seeing the common streams mostly around productivity, but how do you measure that productivity? If it's growth, how are you sort of tracking that growth? And what else is changing? There's a fear I always have when you look at it just from a technology perspective and you're trying to enable something, you may be enabling and scaling a process that itself was broken to begin with. So you end up with these phrase I'll use is you run the risk of having AI-powered Rube Goldberg machines across your organization. So you do want to step back and hone in, what am I really trying to do here? What am I trying to change here? And it's also a very rapidly changing space.

If you just go back six months, the way we were talking about AI is very different than the way we're talking about today and how we'll be talking about it six months from now. So you also don't want to go all in on a fast moving target like that.

Kelly Kohlleffel (12:34)

Are there some use cases? Are there some particular areas that stand out to you, especially around GenAI that seem to hold promise and that you would be very interested in exploring in the future?

Moin Haque (12:46)

Sure. I think there's quite a few things and I would say not just generative, there's so much that hasn't been tapped with traditional or foundational AI because it comes down to what can I scale? What are the patterns I can unlock? What are the insights I can draw at scale through some of those foundational AI capabilities? And generative introduces new things as well because now I can generate scenarios. You'll hear a lot of times in manufacturing the idea of really creating well-formed digital twins, whether it's algorithmically, whether it's spatially, to be able to scale and run scenarios and be able to even have this idea of how do I posture for potential failures and risks. So there's a lot there, but I think it's not just limited to generative. We haven't really explored or taken advantage of all the foundational AI capabilities.

Kelly Kohlleffel (13:40)

How do you balance that need to constantly innovate, you know, kind of certainly, I guess, not get left behind? You've got to maintain security governance. You talked about, you know, that risk side, that safety side, without risking not only the data quality side or governance around data, but even the physical. 

Moin Haque (14:04)

I think it goes back to having the right foundation. The term we like to use is the right guardrails. And they're not there to block. They're there more to guide. And that's what I'm seeing work. That's what we sort of are trying to advocate for. There's this idea like in the term we use is trust councils. It's not governing bodies. They're trust councils. Bring together the right folks, folks across your risk audit teams, your information security teams, your legal privacy teams. Bring the contextual folks, operations, functions. And the trust council really is where all this comes together and says, okay, here's how we can do this in a way that doesn't impede us, but also protects us. And how do we translate that into the hands of those who are actually building things, whether it's product teams or other types of teams. Often when, you know, the analogy I use is if you have a software development team or product team in-house, you'll typically have someone like an enterprise architect or an experience architect, who is again bringing concepts and principles and translating them, putting it in the water of the way that team is ideating. Similarly, have a compliance or safety or privacy champion or architect being part of that as well, where they can contextualize and say, this is how this relates to what you're trying to do. So it's not something you check after the fact, but you sort of do it along the way.

Kelly Kohlleffel (15:34)

That's, that's very, very interesting. How do you go about really putting this together, the coalition of the willing for this trust council?

Moin Haque (15:43)

I think it's going to depend on the cultural specifics for any organization. It's going to depend on the performance maturities across the teams. Again, depending on where an organization sits, it could be something that's stewarded top down, or it could be something that you really need to sort of have a groundswell for and bring together organically. And then do small efforts to show the value, to show the wins. And those wins can be proxies to really get other folks convinced that this does unlock it better than the way we were doing it before. 

Kelly Kohlleffel (16:14)

Got it. Got it. I want to transition here. When you think about your experience, the leadership that you have brought to data teams and organizations in general, what qualities for you, for you personally, or that you see in general, are most important when you're leading a team, any type of team.

Moin Haque (16:35)

So I do get asked often with teams I steward, especially when new folks are coming on the team or transitioning in the team is what are the skills? What are the capabilities I should focus on? And for me, I've sort of come to three that for me are sort of foundational. The first one always is around empathy.

Going back to the other food ingredient example we were talking about, it's that shared table on which we all engage on. And using empathy as that lens through which we can relate to each other and contextualize a situation, that really becomes a guiding skill set and one that I sort of put at the top of the list. It's how I perceive a situation, it's how I posture for it, it's how I respond to it. And in that posture, empathy, typically moves me towards leaning in, which I always prefer versus leaning back in a scenario. 

The second one is entropy. Entropy sounds weird when you mention it to folks, but the thing is, what I found successful is that you have to be able to revel in chaos, essentially be adaptive. We're not always going to have a fixed aperture. We're not always going to have a plan that's well laid out. So you have to be able to respond and adjust accordingly, especially in the data space. We're not building skyscrapers or bridges here. It's not going to be well orchestrated. It's going to be a lot of adaptation along the way to be able to have that ability to be flexible, adaptable, and not be discouraged by entropy, by those changes, is going to be key. Because we're sort of all doing a dance here, and there's too many exogenous shocks. So entropy helps us not just, you know, we're leaning in already, but this keeps us on our toes so to speak. 

And the third one is efficacy. And to me, there's a really important nuance between efficacy and efficiency. And I like efficacy more because it lets me be more holistic, but also realistic around the outcomes to help me make sure that I'm getting to where I'm getting to. I'm dancing towards where I need to dance to. It's not so much about dancing perfectly, but it's good enough. So those, I think, three skills to me are always at the top of the list. 

Obviously, you need the technical skills, you need the critical reasoning, the critical thinking, but technology will keep changing. It's these sort of foundational ones that keep us sort of well-rooted.

We often, when we start with technology, find ourselves going towards a solution. And if we can fall in love with the problem and dwell longer with the problem, I think that actually helps us towards a more sustainable solution.

Kelly Kohlleffel (19:14)

That's fantastic. Looking ahead into next year, are there any top one, two or three key trends that maybe you see advancing? We possibly would touch on a couple already over the next, say 12 months.

Moin Haque (19:30)

That's a pretty long timeframe these days to try to predict for. But I think the thing I think will continue is what we've sort of seen this year, which is that there's a material shift in the definition of what we call data. That data center of gravity has shifted. It's not so much the platforms or the productivity suites or the structured systems. It's all the other repositories of organizational knowledge that can now be unlocked.

And that really has sort of shifted the scale. And what that's also exposed is that a lot of those signals, those repositories, they were never postured for discovery. They were never postured for metadata. They were never postured for governance. Think of the files one would put out on a shared drive or the thread of chats that you're having with folks or the emails. But that's the knowledge that's now accessible. And that was unlocked very materially in the past year.

So the shift with this new definition or scope of data is how do we start to actually leverage and drive value from that? And I think what's going to start competing with air time, the way GenAI and so forth has consumed the air time this year, we're going to start to see more and more us going back and talking about knowledge graphs, talking about taxonomies, talking about ontologies. So much so that I do look forward to getting tired of those terms next year, but I think we need to have those conversations. 

So really seeing that come into the landscape around data is going to be key. My personal interest in that is that anything that gets us away from reports and dashboards, I'm a huge fan of because that is the least, from my perspective, the least efficient way to drive insights. But I do see that becoming a big area of conversation.

Kelly Kohlleffel (21:21)

Fantastic. Moin, this has been exceptionally insightful. I have learned a ton. I just want to thank you so much for joining the show today. I really, really appreciate it.

Moin Haque (21:34)

Same here. Thank you, This was my pleasure.

Kelly Kohlleffel (21:36)

Absolutely. Looking forward to keeping up with everything that you're doing at IFF. And I'd also like to just give a huge thank you to everyone who listened in today.

We appreciate each one of you. We'd encourage you to subscribe to the podcast at any of the major platforms. You can go to Spotify, Apple, Google. You can find us also on YouTube and visit us at Fivetran.com/podcast. Also, please always send us any feedback or comments at podcast@fivetran.com. We'd love to hear from you soon. Take care.

‍