 From the CUBE studios in Palo Alto in Boston, connecting with thought leaders all around the world, this is a CUBE conversation. Hi everyone, this is Dave Vellante and welcome back to the CUBE's coverage of the MIT CDO IQ. This is, God, we've been covering this show since probably 2013, really trying to understand the intersection of data and organizations and data quality and how that's evolved over time. And with me to discuss these issues is Krishna Chariath who's the Vice President and Chief Data Officer, Bristol Meyers Squibb. Krishna, great to see you. Thanks so much for coming on. Thank you so much, Dave, for the invite and looking forward to it. Yeah, first of all, how are things in your part of the world? You're in New Jersey, I'm also on the East Coast. How are you guys making out? Yeah, I think it is, these are unprecedented times all around the globe and whether it is from a company perspective or a personal standpoint, it is how do you manage your life? How do you manage your work in these unprecedented COVID 19 times has been a very interesting challenge. And to me, what is most amazing has been, I've seen humanity rise up. And so too, our company has the snap to be able to manage our work so that the important medicines that have to be delivered to our patients are delivered on time. So really proud about how we have done as a company and of course, personally, it has been an interesting journey with my kids from college, remote learning, wife working from home. So very lucky and blessed to be safe and healthy at this time. So hopefully the people listening to this conversation are finding that they are able to manage through their lives as well. I am, obviously, Bristol-Maris Squibb, very, very strong business. You guys just recently announced your quarter. There's a biologics facility near me in Devons, Massachusetts, I drive by it all the time. Beautiful facility, actually. But extremely broad portfolio. Obviously some COVID impact, but you're managing through that very, very well. If I understand it correctly, you're taking a collaborative approach to a COVID vaccine. You're now bringing people physically back to work. You've been very planful about that. My question is from your standpoint, what role did you play in that whole COVID response and what role did the data play? Yeah, I think it's a two-part. As you rightly pointed out, Bristol-Maris Squibb, we have been an active partner in the overall scientific ecosystem supporting many different targets that is from many different companies. I think across biopharmaceuticals, there's been a healthy convergence of scientific innovation to see how can we solve this together. And Bristol-Maris Squibb have been an active participant as our CEO as well as our chief medical officer and head of research have articulated publicly. Within the company itself, from a data and technology standpoint, data and digital is core to the response from a company standpoint to the COVID-19. How do we ensure that our work continues when the entire global workforce pivots to a kind of a remote setting? So that really calls on the digital infrastructure to the rise to the challenge to enable a complete global workforce. And when I mean workforce, it is not just the employees of the company, but all of the third-party partners and others that we work with. The whole ecosystem need to work. And I think our digital infrastructure has proven to be extremely resilient in that. From a data perspective, I think it is two-fold. One is how does the core book of business of data continue to drive forward to make sure that our company's key priorities are being advanced? Secondarily, we've been partnering with our research and development organization as well as medical organization to look at what kind of real-world data insights can really help in answering the many questions around COVID-19. So I think it is two-fold. Main summary of one is how do we ensure that the data and digital infrastructure of the company continues to operate in a way that allows us to progress the company's mission even during a time when globally we have been switched to a remote working force except for some essential staff from lab and manufacturing standpoint. And secondly is how do we look at the real-world evidence as well as the scientific data to be a good partner with other companies to look at progressing the societal innovations needed for this? I think it's a really prudent approach because let's face it, sometimes one-shot-all vaccine can be like a playing roulette. So you guys are both managing your risk and just as I say, financially a very, very successful company in a sound approach. I want to ask you about your organization. We've interviewed many, many chief data officers over the years and there seems to be some fuzziness as to the organizational structure. It's very clear with you, you report into the CIO. You came out of a technical background. You have a technical degree, but you also, of course, have a business degree. So you got, you know, you're dangerous from that standpoint. You got both sides, which is critical, I would think, in your role, but let's start with the organizational reporting structure. How did that come about and what are the benefits of reporting into the CIO? Yeah, I think the genesis for that as Bristol Maya Squibb, and when I say Bristol Maya Squibb, we are, the new Bristol Maya Squibb is a combination of heritage, Bristol Maya Squibb and heritage cell gene after the cell gene acquisition last November. So in the heritage, Bristol Maya Squibb pre-acquisition, we came to a conclusion that in order for BMS to be able to fully capitalize on our scientific innovation potential as well as to drive data-driven decisions across the company, having a robust data agenda is key. Now the question is, how do you progress that? Historically, we had approached a very decentralized mechanism that made a different data constituencies. We didn't have a formal role of a chief data officer up until 2018 or so. So coming from that realization that we need to have an effective data agenda to drive forward the necessary, data-driven innovations from an analytic standpoint and equally importantly, from optimizing our execution, we came to a conclusion that we need an enterprise-level data organization. We need to have a first-time equals, if you will, to be mandated by the CEO, his leadership team, to be the kind of an orchestrator of a data agenda for the company because a data agenda cannot be done individually by a singular CDO. It has to be done in partnership with many stakeholders, business, technology, analytics, et cetera. So from that came this notion that we need an enterprise-wide data organization. So we started there. So for a while I would joke around that I had all of the accountabilities of the CDO without the lofty title. So this journey started around 2016 where we created an enterprise-wide data organization. And we made a very conscious choice of separating the data organization from analytics. And the reason we did that is when we look at the world of Priscilla Maya Square, analytics, for example, is core and part of our scientific discovery process, research or clinical development, all of them have deep data science and analytics embedded in it. But we also have other analytics, whether it is part of our sales and marketing, whether it is part of our finance and the enabling functions, we catch all across global procurement, et cetera. So the world of analytics is very broad. BMS did a separation between the world of analytics and from the world of data. Analytics at BMS is in two modes. There is a central analytics organization called Business Insights and Analytics that drive most of the enterprise-level analytics. But then we have embedded analytics in our business areas, which is research and development, manufacturing and supply chain, et cetera, to drive what needs to be closer to the business area. And the reason for separating that out and having a separate data organization is that none of these analytic aspirations or the business aspirations from data will be met if the world of data is... We don't have the right level of data available. The velocity of data is not appropriate for the use cases. The quality of data is not great or the control of the data so that we are using the data for the right intent, meeting the compliance and regulatory expectations around the data is met. So that's why we separated out the data world from the analytics world, which is a limit of a unique construct for us compared to what we see generally in the world of CDOs. And from that standpoint, then the decision was taken to make that report for global CIO. At Bristol-Mysk Group, we have a very strong CIO organization and IT organization. When I say strong, it is from this land standpoint. A, it is centralized and we have centralized the budget as well as we have centralized the execution across the enterprise. And the CDO reporting to the CIO with that data specific agenda has a lot of value in being able to connect the world of data with the world of technology. So at BMS, the chief data officer organization is a combination of traditional CDO type accountabilities like data risk management, data governance, data stewardship, but also all of the related technologies around mass data management, data lake, data and analytic engineering and a nascent AI data and technology lab. So that construct allows us to be a true enterprise horizontal supporting analytics, whether it is done in a central analytics organization or embedded analytic teams in the business area, but also equally importantly focus on the world of data from operational execution standpoint, how do we optimize data to drive operational effectiveness? So that's the construct that we have that CDO reports to the CIO, data organization separated from analytics to really focus around the availability, velocity, quality and control of data. And the last nuance that is that at BMS, the chief data officer organization is also accountable to be the data protection office. So we orchestrate and facilitate all privacy related actions across because that allows us to make sure that all personal data that is collected, managed and consumed meets all of the various privacy standards across the world, as well as our own commitments as a company from a trust and transparency standpoint. So that makes a lot of sense to me and thank you for that description. You're not getting in the way of R&D and the scientists, the scientists, they know data science, they don't need really your help, right? I mean, they need to innovate at their own pace, but the balance of the business really does need your innovation and that's really where it seems like your focus. You mentioned master data management, data lakes, data engineering, et cetera. So your responsibility is for that enterprise data life cycle to support the business side of things. And I wonder if you could talk a little bit about that and how that's evolved. I mean, a lot has changed from the old days of data warehouse and cumbersome ETL and you mentioned, as I say, data lakes. Many of those have been challenging, expensive, slow, but now we're entering this era of cloud, real time, a lot of machine intelligence and I wonder if you could talk about the changes there and how you're looking at and thinking about the data life cycle and accelerating the time to insights. Yeah, I think the way we think about it as we as an organization in our strategy and tactics think of this as a data supply chain. The supply chain of data to drive business value, whether it is through insights and analytics or through operational execution. So that's, when you think about it from that standpoint, then we need to get many elements of that into an effective stage. This could be the technologies that is part of that data supply chain. You referenced some of them, the master data management platforms, data lake platforms, the analytics and reporting capabilities and business intelligence capabilities that plug into a data backbone, which is that, I would say, the technology swim lane that needs to get right. Along with that, what we also need to get right for that effective data supply chain is that data layer. That is, how do you make sure that there is the right the data navigation capability? How do you make sure that we have the right ontology mapping and the understanding around the data? How do we have data navigation? It is something that we have invested very heavily in. So imagine a new employee joining BMS. Any organization our size has a pretty wide technology ecosystem and data ecosystem. How do you navigate that? How do you find the data? Data discovery has been a key focus for us. So for an effective data supply chain, then we knew that and we have instituted our roadmap to make sure that we have a robust technology orchestration of it. But equally important is an effective data operations orchestration of it. Both needs to go hand in hand for us to be able to make sure that the supply chain is effective from a business use case and analytic use case standpoint. So that has led us to on a journey to honor from a cloud perspective. Since you've referred that in your question is we have invested very heavily to move from very disparate set of data ecosystems to a more converged cloud based data backbone. That has been a big focus at BMS since 2016. And we have whether it is from a research and development standpoint or from commercialization, which is our word for sales and marketing or manufacturing and supply chain and HR, et cetera. How do we create a converged data backbone that allows us to use that data as a resource to drive many different consumption patterns? Because when you imagine an enterprise of our size, we have many different consumers of the data. So those consumers have different consumption needs. You have deep data science population who just needs access to the data. And they have data science platforms, but they are advanced programmers as well. To the other end of the spectrum where executives need pre-packaged KPA. So the effective orchestration of the data ecosystem at BMS through a data supply chain and a data backbone does a couple of things for us. One, it drives the productivity of our data consumers, be it scientific researchers, analytic community, or other operational staff. And second, in a world where we need to make sure that the data consumption upholds ethical standards as well as privacy and other regulatory expectations, we are able to build into our system and process the necessary controls to make sure that the consumption and the use of data meets our highest trust and transparency standards. Yeah, that makes a lot of sense. I mean, converging your data like that, people always talk about stovepipes. I know it's kind of a bromide, but it's true and allows you to sort of inject consistent policies. What about automation? How has that affected your data pipeline in the last, you know, recently and on your journey with things like data classification and the like? Yeah, I think in pursuing a broad data automation journey, one of the things that we did was to operate at two different speed points. You know, historically, the data organizations have been riddled with long running data infrastructure programs. By the time you complete them, the business context have moved on and the organizational leaders are also exhausted from having to wait from these massive programs to reach its full potential. So what we did very intentionally from our data automation journey is to organize ourselves in two speed dimensions. First, a concept called Rapid Data Lab. The idea is that, you know, recognizing the reality that the data is not well automated and orchestrated today. We need a SWAT team of data engineers, data SMAs to be able to partner with consumers of data to make sure that we can make effective data supply chain decisions here and now and enable the business to answer questions of today. Simultaneously, in a longer time horizon, we need to do the necessary work of moving the data automation to a better footprint. So our enterprise data lake investments where we built services based on and we had chosen AWS as the key backbone for our cloud, as the cloud backbone for data. So how do we use the AWS services? How do we wrap around it with the necessary capabilities so that we have a consistent reference and technical architecture to drive the many different functional journeys? So we organized ourselves in two speed dimensions. The Rapid Data Lab teams focus around partnering with the consumers of data to help them with data automation needs here and now. And then a secondary team focused around the convergence of data into a better cloud-based data backbone. So that allowed us to, one, make an impact here and now and deliver value from data to the business here and now. Secondarily, we also learned a lot from actually partnering with consumers of data on what needs to get adjusted over a period of time in our automation journey. It makes sense. Again, that whole notion of converged data, putting data at the core of your business. You brought up AWS. I wonder if I could ask you a question. You don't have to comment on specific vendors, but there's a conversation we have in our community. You have AWS, huge platform, tons of partners, a lot of innovation going on. And you see innovation in areas like the cloud data warehouse or data science tooling, et cetera. All components of that data pipeline. As well, you have AWS with its own tooling around there. So a question we often have in the communities will technologists and technology buyers go for kind of best of breed and cobble together, different services, or would they prefer to have sort of the convenience of a bundled service from an AWS or a Microsoft or a Google or maybe they even go best of breeds for all cloud? Can you comment on that? What's your thinking on that? Yeah, I think especially for organizations our size and breadth, having a converge to convenient all of the above from a single provider does not seem practical and feasible. Because a couple of reasons. One, the heterogeneity of the data, the heterogeneity of consumption of the data. And we are yet to find a single stack provider to who can meet all of the different needs. I think, so I am more in the best of breed camp with a few caveats, a hybrid best of BradyCube. It is important to have a converged data backbone for the enterprise. So whether you invest in a singular cloud or a private cloud or a combination, you need to have a clear intention strategy around where are you gonna house the data and how is the data is gonna be organized. But you could have a lot more flexibility in the consumption of data. So once you have the data converged into, in our case, we converge to turn on AWS based backbone. We allow many different consumptions of the data because I think the analytic and insights layer, data science community within R&D is different from a data science community in the supply chain context. We have a business intelligence needs. We have a catered needs. And then there are other data needs that needs to be funneled into software as service, platforms like the sales forces of the world to be able to drive operational execution as well. So when you look at it from that context, having a hybrid model of best of breed, whether you have a lot more convergence from a data backbone standpoint, but then allow for best of breed from an analytic and consumption of data is more where my harp and my brain is. Yeah, for that, I know a lot of companies will be excited to hear that answer, but I love it because it fosters competition and innovation. Ash, I could talk for you forever, but you made me think of another question, which is around self-serve. On your journey, are you at the point where you can deliver self-serve to the lines of business? Is that something that you're trying to get to? Yeah, I think it is. The self-serve is an absolutely important point because I think the traditional boundaries of what you would consider the classical IT versus the classical business is great. I think there is an important grade in the middle where you have a deep citizen data scientist in the business community who really needs to be able to have access to the data and I have advanced data science and programming skills. So self-serve is important, but in that companies need to be very intentional and very conscious of making sure that you're allowing that self-serve in a safe containment zone because at the end of the day, whether it is a cyber risk or data risk or technology risk, it's all real. So we need to have a balanced approach between promoting whether you call it data democratization or whether you call it self-serve, but you need to balance that with making sure that you're meeting the right risk mitigation strategy standpoint. So that's how been our focus is to say, how do we promote self-serve for the communities that they need self-serve where they have deeper levels of access? How do we set up the right safe zones for those which were with the appropriate mitigation from a cyber risk or data risk or technology risk? Yeah, that's security pieces. Again, you keep bringing up topics that I could talk to you forever on, but I heard on TV the other night, I heard somebody talking about how COVID has affected, you know, because of remote access, affected security. And it's like, hey, give everybody access. That was sort of the initial knee-jerk response. But the example they gave as well, if your parents go out of town and the kid has a party, you may have some people show up that you don't want to show up. And so same issue with, you know, remote working, work from home. You know, clearly you guys have had to pivot to support that, but where does the security, you know, organization fit? Does that report, you know, separate alongside the CIO? Does it report into the CIO? Are they sort of peers of yours? How does that all work? Yeah, I think at Bristol Myoscope we have a chief information security officer who is a peer of mine who also reports to the global CIO. And we, the CDO and the CSO are effective partners and are two sides of the coin in trying to advance a total risk mitigation strategy, whether it is from a cyber risk standpoint, which is the focus of the chief information security officer and whether it is the general data consumption risk and that is the focus from a chief data officer in the risk capacities that I have. And together those are two sides of the coin that the CIO needs to be accountable for. So I think that's how we have orchestrated it because I think it is important in these worlds where you wanna have, you wanna be able to drive data-driven innovation, but you wanna be able to do that in a way that doesn't open the company to unwanted risk exposures as well. And that is always a delicate balancing act because if you index too much on risk and then high levels of security and control, then you could lose productivity. But if you index too much on productivity, collaboration, open access and data, it opens up the company for us. So it is a delicate balance between the two. And that increasingly we're seeing that reporting structure evolve and coalesce. I think it makes a lot of sense. I felt like at some point you had too many seats at the executive leadership table, too many kind of competing agendas. And now we are structure, the CIO is obviously a very important position, I'm sure has a seat at the leadership table, but also has the responsibility for managing that sort of data as an asset versus a liability, which in my view has always been sort of the role of the head of information. I want to ask you, I want to hit the escape key a little bit and ask you about data as a resource. You hear a lot of people talk about data is the new oil. We often say data is more valuable than oil because you can use it. And it doesn't follow the laws of scarcity. You can use data in an infinite number of places. You can only put oil in your car or your house. How do you think about data as a resource today and going forward? Yeah, I think the data as a new oil paradigm, in my opinion, was unhealthy and it prompts different types of conversations around it. I think for certain companies, data is indeed an asset. If you're a company that is focused on information products and data products, and that is core of your business, then of course there is a monetization of data and then data as an asset, just like any other assets on the company's balance sheet. But for many enterprises further their mission, I think considering data as a resource, I think is a better focus. So as a vital resource for the company, you need to make sure that there is an appropriate caring and feeding for it. There is an appropriate management of the resource and an appropriate evolution of the resource. So that's how I would like to consider it. It is a personal end of one perspective that data as a resource that can power the mission of the company, the new products and services. It needs to, I think that's a good, healthy way to look at it. At the center of it though, a lot of strategies, whether people talk about a digital strategy or whether the people talk about a data strategy. What is important is a company to have a true north star around what is the core mission of this company and what is the core strategy of the company. For Bristol Myoscope, we are about transforming patients' lives through science. And we think about digital and data as key value levers and drivers of that strategy. So the digital for the sake of digital or data strategy for the sake of data strategy is meaningless in my opinion. And we need to, we are focused on making sure that how do we make sure that data and digital is an accelerant and has a value lever for the company's mission and company's strategy. So that's why thinking about data as a resource allows us as a key resource for our scientific researchers or a key resource for our manufacturing team or a key resource for our sales and marketing colleagues allows us to think about the actions and the strategies and tactics we need to deploy to make that effective. That makes a lot of sense and you're constantly using that north star as your guide light and how data contributes to that mission. Krishna Charyath, thanks so much for coming on theCUBE and supporting the MIT chief data officer community. It was a really pleasure having you. Thank you so much for Dave. Hopefully you and all the audience stay safe and healthy during this time. Yeah, thank you for that. And thank you for watching everybody. This is Dave Vellante for theCUBE's coverage of the MIT CDO IQ Conference 2020 gone virtual. Keep it right there. We're right back right up for this short break.