Why AI Starts With Effective Data Management

Read Interview

Speaker 1:
Welcome to Champions of Data and AI. Brought to you by Databricks. In each episode, we salute champions of data and AI, the change agents who are shaking up the status quo. These Mavericks, are rethinking how data and AI can enhance the human experience. We’ll dive into their challenges and celebrate their successes all while getting to know these leaders a little more personally.

Chris D’Agostino:
Welcome to episode six of Champions of Data and AI. I’m your host, Chris D’Agostino. Before we get started, I want to thank each of you for joining me and our champions this season. In each of the episodes, we explored the role Data and AI plays in today’s world, and how leaders are approaching their data and AI strategies. This episode will conclude season one. I’ll be back in June for another exciting season, and we’re going to mix it up a bit. I’ve asked Databricks’ very own, Alexandra Mysak to join me as co-host starting in season two. I think you will love hearing her perspective.

Chris D’Agostino:
Now let’s dive into today’s discussion. By now, we all know, in order to have an effective enterprise data and AI strategy, you need to take a methodical approach to data management across the organization. In this episode, I’m joined by Habsah Nordin. She is the head of enterprise data in group digital at Petronas. Habsah discusses her approach to building a data-driven organization. We’ll also get her perspective on her experience as the first woman at Petronas in this role and what advice she has for those aspiring to become data and AI leaders. Let’s get started.

Chris D’Agostino:
So Habsah, thanks for being here today and talking with us.

Habsah Nordin:
Thank you, Chris. Thank you for having me in the Champions of Data and AI.

Chris D’Agostino:
Awesome. Well, so, you know, I know that you’re the GM of enterprise data in the group digital at Petronas. One of the things that, when we last talked, you shared with me that you’ve been with the company pretty much your entire career, and you’ve had different roles within the organization. It struck me that having those different roles really gives you a really good breadth of understanding with how data is being used throughout the organization. And so, as you talked about data transformation and how the company is trying to transform itself, I found it interesting that you’re the first female leader in the data space. And so we’d love to hear your thoughts on that and what that means for Petronas as you move forward.

Habsah Nordin:
Yeah. Thank you, Chris. I think that I have actually spent almost 25 years in Petronas, so I don’t really have profound knowledge in data management. My recent involvement in data only started in 2019. I just started to read the enterprise data in Apple 2020. But over the years that I’ve worked in other parts of the non-IT department, I have been in the strategic planning role, the internal audit, the marketing, the project management, and I can understand exactly the pin points of having data, bringing data and analyzing data.

Habsah Nordin:
And I think the more important is having that understanding of perspective external in the enterprise data. How do we see data as one of the key in that being transformation of digital? To me, there is very key where we look at the data space because making the data available will be the first thing that we need to do right before we can able to monetize the value of data.

Habsah Nordin:
So, Chris, I think being here in enterprise data, I would say is a privilege because I think being a part of the transformation journey of digital would be one of the milestone that not just for myself and my teams are creating, but also for the entire Petronas. I think it’s not about me being the first female leader. But in Petronas, we practice about high-performance culture. It’s all about the merits and having the right talent to actually move these transformation or digital forward. And data will be one of the key enabler.

Chris D’Agostino:
Yeah, that’s great. Can you help our listeners better understand Petronas and some of the work that you’re doing? Many of our listeners probably don’t realize just how big of an organization you are and how many locations you operate in. So why don’t you educate us a bit on the role of the company in the oil and gas industry?

Habsah Nordin:
Okay, Chris, so to be… Petronas, actually, is a national oil company in Malaysia. We have been formed since August 1974. And we have had 325 subsidiaries that we have been operating in more than 65 countries around the globe. We are now employing nearly 248,000 employees. So as an oil and gas company, we are a very highly integrated oil and gas value chain from exploration all the way into marketing. So when you talk about from the business angle, and you define that to a data space, it is a very white space and is highly complex.

Habsah Nordin:
If you take a look at each of the data domain in relation to the business, I think people would understand that there are specific… and that I would say, unique issues that we face from a context of data in each of that business domain. So this is exactly how the complexity arises, not just from a context of just running the business, but even making an… The data we make available across Petronas is one of the key priorities that we are now moving ahead with enterprise data.

Chris D’Agostino:
So when you think about those priorities, what are the ones that are the challenging ones… as we like to say in industry, the things that keep you up at night? And then, give us some examples of problems that you’ve solved that you’re really proud of. I know you’re relatively new in the role and you also don’t come from the data space. So I’m curious about things that you might’ve been able to accomplish that perhaps even on a personal level, surprised you and you’re proud of and would like to share.

Habsah Nordin:
Yeah. So I think first and foremost, the privilege that I have is I have a team. It’s a new setup team that we actually bring on-board from within Petronas as well as we bring new talents from outside. And what we are now embarking will be the first pioneering journey to build a unified data platform. We call it enterprise data hub. So this is a historical project for us, but in order for us to do this right, the first thing we need to do is to fix the data availability. What it means is that, your analysis, we are retuning for the right process and governance people across Petronas, grow accompanies so that it allows for a seamless and effortless data liberation of Petronas both in vertical, as well as horizontal sharing of data. So we need to do this, obviously, that we start from a very strong commitment and the right tone from the top. And in order to do that, that will be first thing that we need to do it right… bringing the whole enterprise data hub to life.

Habsah Nordin:
The second part about it is also about the change management approach. We need to have a strong narrative. Why do we need to liberate data? Because I think, Petronas in the past, I think we have been very much inclusive where we protect our own data within each of the companies. So we need to have the ability to communicate why we should not be having fear about liberating and sharing data across Petronas’ environment. And this is itself a new pervasive for basic culture actually for Petronas.

Habsah Nordin:
One of the thing that we see from the enterprise data hub is one of the key foundation that will power up many of the digital use cases, including the digital twin agenda that we are now aggressively looking at. An enterprise data hub will be an integral part to unify data that comes from various data platforms in Petronas across all data sources in the ecosystem. So it allows for EDH to be functioned as single source of truth.

Chris D’Agostino:
Well, I was going to ask, when you think about combining data from across organizations, there’s… at least in my experience having done this type of work in the past, as well as now in my role at Databricks, talking to other leaders like yourself, and working with some of the larger companies in our portfolio of customers, that combining those data sets are oftentimes the breakthroughs to new insights for the company, and ways in which you can increase revenue root, reduce risk, reduce your costs, but it requires that sort of business leaders and technology leaders partner together to share that data, and you have to, from a data governance standpoint, really guarantee and ensure that the data quality is kept very high, and that the data governance and lifecycle management is there. So can you tell us a little bit about how, with the enterprise data hub, are you approaching that?

Habsah Nordin:
Yeah, that’s true Chris. I think you actually pointed out what exactly are the critical point of success for enterprise data hub to become the single source of truth. At the moment what we are doing is putting up and laying out the foundation right. The first thing is, how do we define the right data roles in the organization and making the right person that the people, be able to be accountable and carry on that duty as the data role that we define. That’s the first thing.

Habsah Nordin:
The second thing is we are not embarking into the enterprise data quality programs. What it means is that we actually bring the right tools. We are actually putting up the right process, up-skilling the people and having the right monitoring precise on the data quality.

Habsah Nordin:
To ensure that the data be able to monetize the value, the first thing you need to do right, is to trust the data. So I think what a lot of organizations are struggling with is how do we ensure a consistent data quality that we can trust. This is very important for us in Petronas. And I think it’s more important soon that we’re going to embark… I wouldn’t say embark, but we’re going to be more pervasive and more aggressive in the AI journey. So having to have data quality is important. In order to do that as well, I think we are also looking at having the right data standards, because at the moment, when we look at most of the data, the common issue that we are facing right now is the data in the operability. Most of the time, the data is actually being clocked within the propriety applications, right? Because you don’t really have a data standard that we able to decouple data from the applications. So that is tough.

Habsah Nordin:
It’s also one of the key program that we are now doing in Petronas to actually adopt the industry data standards. We also define Petronas data standards and be able to apply in adopt this across the whole application. And more so, it becomes a technical gut roof that we define for any new applications that we’re going to bring on board. So we are actually fixing what I call is, the issue of the day, as well as having a better vision for tomorrow. So this is how looking at the data standard, data architecture, as well as data quality, play an important part of ensuring that this is well done and done in an integrated manner.

Chris D’Agostino:
Well it speaks to the core philosophy of how Databricks has built its platform, right? We have this platform, this approach to how you work with data, we call it lake house. The concept there is to take the benefits of data governance that you see in an enterprise data warehouses, take the flexibility that you get with your data lakes, and be able to combine the best of both worlds, where instead of having data sets locked in these source systems in different formats, with different structure, that to really enable machine learning at scale, you actually need to give your data science and machine learning experts access to more of the data.

Chris D’Agostino:
But the data is, as you say, it’s got to be of good quality. So our approach is, we want to use this lake house concept as a way to bring data in from all these source systems, put it into one single source of truth, which is like your cloud-based object store. So your Azure ATLS, or AWS S3, since I know those are the platforms your company runs on, and be able to curate that data and force the schemas, do the data quality checks in one place, refine the data, normalized date time stamps, things that you need to be able to do to get cross business unit data sets to work together and actually be meaningful. The platform is really designed to enable that across different personas, whether you’re a Python developer, SQL developers or Java or Scala developers.

Habsah Nordin:
So that’s the reason why we have actually identified Databricks, one of the technology stack that we are going to build within the EDH. That allows us to actually bring data and connect to that across different data sets, across all the applications and data platforms.

Habsah Nordin:
Additionally to that, what’s important that we need to do is to have the right design of the data model within the data platforms. That’s why we need to actually have, how do we design the common plus industry data model, and be able to converge that with the data model that we generate from the advanced analytics. This itself is a new, as well as a pioneering area, where we push the boundary of thinking. We have a team that believes in themselves that we can do this right, and working with the right partner, delivering EDH to his full potential by end of 2021.

Chris D’Agostino:
Awesome. Let’s shift gears a little bit because as you know, part of this is a technology solution, but oftentimes it’s very much a cultural change. And it’s getting the organization to think about data in ways that it hadn’t thought about data before. And it’s really kind of moving more towards thinking about data as an asset, rather than something that’s in a given source system, a fit for purpose store, and for use cases that apply to that line of business or that department within a line of business. It’s now thinking about data much more broadly. So can you talk a little bit about steps that you’ve taken to create a more pervasive data culture within Petronas?

Habsah Nordin:
Yeah. Chris. Yeah, this narrative about data as an asset, this has been the very constant narrative in Petronas. And we have seen how, when we start to embark on the digital projects why… what we have seen in terms of the data that we have explored, and the new insights that we have generated that it is a belief system that we are now building within Petronas.

Habsah Nordin:
I think more important in terms of the culture is, while we are making this narrative more prevailing, we are also making it evident through the digital projects that we are now progressively undertaking. And we are also looking at having machine learning. And we have uplift that, and has many successful projects in that space as well. And that makes people excited… That makes people excited that there are now belief for a data as an asset. They are now taking more responsibility to be accountable on the quality of the data. That itself I thing is a mock of first revelation that data is an asset.

Habsah Nordin:
The second part is about, we are also embarking on the citizen analytic program. This program is actually to level up the employees with… to make sure that we are more analytic savvy. So actually that helps with regards to having the undertaking, understanding about how do we approach analytics and having the right data being available. That allows for many parts of employees in this organization to draw new insights. With the new insights, we know that there will be a lot of plenty new thinking, right? And the ability to take different actions altogether. We have seen this in many pockets of success in Petronas. So I think where the culture is now moving on is very pervasive itself, that people talk about the value of data. And it is very, very evident and it’s quite pervasive across betterness.

Chris D’Agostino:
Yeah, that’s, that’s great to hear. So it sounds like if you were giving advice to other leaders who were beginning their journey, the notion of finding more value, more insights in the data by collecting more of the data, getting it into a single source of truth, ensuring that you’re up-skilling the employees in the organization so that they can build their skills to analyze that data, and unlock some new insights and maybe progress their skill sets, progress their careers… These are the sort of inherent benefits of moving towards this type of data architecture and data strategy. Anything else that you’d… What advice you would give to other leaders that need to be able to do this within their organizations?

Habsah Nordin:
Yeah, that’s true Chris. I think when we started, we went with the journey of data. The first thing we understand about is to be clear on the value of the data. So it’s not about bringing the quantity, but having the right data that brings the right value and ensuring that the whole organization will be accountable in the quality of the data. But I think what is more important too is, even building the data pipelines itself is a cost, right? How do we shift that as a return on investment approach rather than people thinking about it as a cost to invest? So I think there’s also a very dynamic team and that allows us to ensure that we actually put in the resources in the right assets, because I think data itself is not the pin-points of data. I think it is quite already across the organization, right?

Habsah Nordin:
I think many organizations face this problem about… the issue about stranded data, the issue on interoperability on the quality of data, but where do you start? So you have to start when you prove when there is that proven use case, that proven a value to data. So the whole missionary in the core assets is actually to ensure that we have the right tracks and programs to support that particular use cases.

Habsah Nordin:
So I think that’s more important. It’s more of a targeted approach in managing data. I think at the same time as well, Chris, there would also be some of the team ethic programs that I think the organization may, will need to undertake. This is also what we have done in Petronas. It is as simple as we talk about bringing the right data standards and data quality. We do that with involvement of the people, the subject matter experts in the business. So there are things that within the control where as the enterprise data in my department, that we take role, but there are certain things that we bring in also the expertise of business to help us with the data journey. So I think there will be some of the lessons learned, as well as some of the advisers that I could provide to the other leaders in terms of the data space that you wanted to pursue in your organization.

Chris D’Agostino:
So Habsah, whenever I’m talking to one of the leaders that is helping transform data and AI inside their organization, I often ask them, what piece of advice they would give to someone trying to build a career in this space. And what’s fascinating to me about you is that in organizations, I see oftentimes that you have very, very strong technologists that understand the tooling and the very specifics about machine learning algorithms and things like that. But they oftentimes don’t have a detailed knowledge of the data and how the business uses the data.

Chris D’Agostino:
And then I go over and I look at the business side of the house and they will really understand what data sets they have, what value they’d like to get out of the data if only they had the right tooling or technology to help them do that. And so your background is really interesting because you’ve got a technical background, but you’re new to the data space and the tooling and data specifically, but you’ve had experience across a wide range of areas of a global company. And so I’m just curious how you see that as both a challenge for you in terms of you and your team getting deeper in the tech, but also… maybe have a little bit more momentum because you actually understand how the business operates with data today.

Habsah Nordin:
Okay, Chris, I think going back to the questions, right? I think whether you are more profound in the business versus profound in the technology, to meet these are the balancing of dynamics that we need to have. Take me, for example. What I can add value to the team and with the data journey is… My level of experience is involvement in the non-data space, which is the business place, right? So I understand what exactly are the business needed. When I’m here in enterprise data, I explore new areas. And that area is about the technology. What exactly are the tools, technology, that helps in terms of harnessing the real value of the data. That itself is where the journey of learning for myself as well as the team. And the thing is when we talk about data, it’s a blend of both. It’s not just a vertical and I would say, inclusion with respect to the area.

Habsah Nordin:
It has to be married between the two. I think the conversation that we need to have more, it’s not really from a technology sense because I think that analogy come and go, it will be developed over time and there will be improvement and modernization, and new functionality.

Habsah Nordin:
But then what’s more important is, what do we get from the data that we wanted to further harness, we wanted to further monetize? And the technology would just be the enabling function to be able to draw upon that ability. So, going back to that notion, I think it there’s no right and wrong, but I think it’s just that in my case, I would then have to be more in-tune and more focused, and build my understanding and agility in the technology space, and bringing that business perspective to be able to draw upon the right strategy and as well as the right priorities. One thing also to notice is not about how sophisticated that you have with your technology, because you could do that from an investment standpoint, but what do you get from that knowledge investment, how much value are you getting from the data that you bring in and for the use from the tunnel? I think that is what really matters as many of innovation in the space of data agenda.
Speaker 1:
Thank you for joining this episode of Champions of Data and AI. Brought to you by Data bricks. Thousands of data leaders rely on Data bricks to simplify data and AI. So data teams can innovate faster and solve the world’s toughest problems. Visit databricks.com to learn how data leaders are unlocking the true potential of all their data.