Skip to content

Search the site

The Big Interview: GSK Chief Digital and Technology Officer Shobie Ramakrishnan

"We have a federated Enterprise Architecture approach; making sure that we don’t get constantly caught in these endless loops of centralisation and decentralisation..."

At the start of her career, Shobie Ramakrishnan spent seven formative years at Apple, helping it modernise its IT landscape as the company returned to growth and profitability, joining as an applications development leader in 1996; the same year that legendary CEO Steve Jobs returned to shake up the company. Two things about her experience there left a powerful impression, and have informed her career and leadership since – including in her current role as a change-making Chief Digital and Technology Officer (CDTO) at the UK’s pharmaceutical powerhouse GSK. 

She recalls: “It was very clear that we were working on something that was way bigger than ourselves. We really believed that we were creating incredible capabilities that were going to power up the best, most talented people in the world to achieve the things that they needed to achieve. That was very palpable. People think of Apple as a design company; everything is so beautiful and well architected and connected and integrated and all of that. But in my opinion, the best gift that Jobs gave to Apple was focus – to say 'you're working on these things; not on these things'"she adds.

Both this sense of mission and focus inform Ramakrishnan’s work as CDTO. She joined GSK in 2018 after a career spanning pharmaceuticals (at both AstraZeneca and Genentech) and in enterprise software (Salesforce) characterised by reform and significant M&A integrations; for example leading a technology integration between Roche and Genentech after a $46.8 billion buyout and then Salesforce's $2.5 billion ExactTarget tuck-in.

So what is the current mission?

An inspiring mission… 

“We are a company that is founded on and built on the mission of being able to develop transformational medicines and vaccines; that's who we are: our technology is in service of that mission” she says to The Stack – emphasising that for a digital and technology leader, pharmaceuticals is one of the single most exciting sectors to be in, as technological innovation fundamentally transforms and powers how that mission is achieved. 

(As GSK's Chief Scientific Officer Tony Wood puts it: "There’s never been a time when science and technology have formed such a powerful union as they do now. We’re able to identify drug discovery targets using genomics and AI/ML in a way that is unprecedented. We’re also able to use more approaches than ever before to design new vaccines and medicines...")

That’s the mission. Her focus as CDTO is fourfold, Ramakrishan says – sitting down to chat with The Stack, shortly before GSK reported its third-quarter earnings showing that it had exceeded its full-year guidance expectations (hitting £22+ billion in revenues so far this year.) 

Four areas of focus

The first is developing capabilities to accelerate GSK's product pipeline, the second is reimagining the supply chain and its manufacturing processes, the third is supporting commercial execution and scientific customer engagement, and the fourth is ongoing work to improve how GSK uses data and analytics across lines of business, to drive agile decision-making, improve efficiencies, and foster innovation.

Discussing those points, we also touch on GSK’s approach to cloud, engaging the startup ecosystem, breaking down silos across business, technology, science; Ramakrishnan’s uniquely fluid-sounding approach to how she works with lines of business – and some powerful leadership lessons learned during her career in Silicon Valley, Europe and beyond. 

Priority 1: Product pipeline

“We are living through an unprecedented time when it comes to the nexus of biology and technology” says GSK’s CDTO. “It’s almost a cliché now, with everybody talking about AI and drug discovery, but over the next two decades we will see the evolution and iteration of these technologies and it will drive a paradigm shift in scientific discovery. 

She says: “We have been preparing and building capabilities towards this future for the past 5+ years. We’ve built incredible data capabilities, with our ability to generate and use genomic data; with our scientific AI/ML capabilities in research, that helps us make sense of the enormous amounts of data that we generate and that we have access to from external sources –and we have proof points that our strategy is working.” 

“For example, our research team tells us that there has been a two-times increase in the number of targets from CRISPR screens and a six-month reduction in cycle times.” (CRISPR screens are a source of biological discovery that allow interrogation of gene function. Identification of small molecules, or targets, to decipher their mechanism of action is an important process in drug development. More targets: more success…)

She adds: “We've been able to improve the yield of our clinical supply by 5% to 10%; reduce development cycle time by 50% in those areas where we've adopted digital twins, (e.g.) for a brand new RSV vaccine that we just launched. So we have very specific examples [to showcase] that the strategies are yielding better targets and improving probability of success – we're one of the industries that sort of accepts 90% failure of our ideas as success; only 10% of our ideas make it through the pipeline. But with the help of data and technology, we can really make a dent in that.”

Priority 2: Reimagining the supply chain

GSK is also working to optimise its supply chain; seeing where and how technology can help improve processes and also boost sustainability

“A very important part of our estate and priorities is what our supply chain team is working on,” Ramakrishnan confirms. “We're looking at all aspects of this lifecycle: internal and external supply chains, manufacturing and quality, the incredibly complex and regulated, but critical aspect of process transfer of our medicines as we scale from clinical supply to large-scale manufacturing to get it ready for market launch.”

She adds: “This is an incredibly tough area in which to drive big technology change – and one that we have to be very careful about as patient safety is paramount. We have the regulatory controls to ensure that we don't move fast and break things. It might sound fun to do in a social media company, but it does not work well in our industry. 

“But it's really rewarding when you do get it right.”

What does this mean, more concretely?

“[Building out] analytics capabilities [in labs] that includes digital twins to drive yield improvements, efficiencies, improved quality and safety, and to create a much more connected and agile supply chain [are priorities].” 

A specific example: “We have emerging evidence that confirms that our strategy to embed data and technology into our problem-solving is delivering results. We've been able to increase our batch yield by 10%, for one of our respiratory medicines by adopting a digital twin that helped us model the best ways to optimise yield,” she tells The Stack.

This is not just a tech challenge of course: “We recognise that it's not just data or technology, that's just one piece of the puzzle. We achieve the outcomes by combining it with traditional business best practices like driving lean processes, or automating our schedule optimization. So the new has to sit alongside the old to really drive these amazing outcomes. 

Priority 3: Scientific and customer engagement 

A huge pillar of anyone’s work at GSK is scientific and customer engagement; another area where improved data capabilities offer promise. There’s a real opportunity to deliver more “data driven insights” and an “incredible opportunity to reach more doctors and therefore more patients” says Ramakrishnan – admitting it’s not all plain sailing.

“The reason it's hard to disrupt an industry like ours – and I would say we are still in the bottom half of industries when it comes to digital disruption – is that transformation is not a one-industry problem.

“The status quo is anchored in a healthcare ecosystem that is hyper-localised to each country that we operate in. In the US, we have the providers, we have the insurance companies, we have life sciences companies, individual doctors, the patient's experience of health care; all of these things coming together in some way, shape, or form. It's not about going to one website to drive your ultimate experience in getting the health care that you need or deserve... So therefore, the data is the fluid that can connect and work across these healthcare ecosystems.” 

Priority 4: Data federation

Using data as a connective tissue is, perhaps, easier internally, but even there, like in any large multinational, large silos have traditionally existed.

This brings us to her fourth priority pillar: Improving how GSK uses data and analytics across lines of business, to drive agile decision-making.

Ramakrishnan says: “We don't have an ecommerce website, or a streaming app, or a banking portal that anchors our customer experience, and then converges the enterprise behind it in a digital strategy like many industries – I think recognising the ecosystem in which you're solving problems is by far the most important skill for a leader; if I was showing up at an Apple or at a Charles Schwab, today, I would be setting different architectural expectations and different strategies than I do at GSK.”

“But data is the golden thread…

A federated approach… 

She says: “We have a federated Enterprise Architecture approach; making sure that we don’t get constantly caught in these endless loops of centralisation and decentralisation that CIOs and CTOs tend to get trapped in; it's exhausting to be on that pendulum!

“This federated approach to AI and data architecture [means you have to do] 20% of your things together – you don’t get a lot of leeway on data platform – but the 80% is what I call deep empowerment to go and create value. If you have good technologists, good engineers,, then you can take a federated approach without losing control of the ecosystem...”

How do you do that without losing control, The Stack asks?

“We allow a bit of freedom and genuine creativity at the top of the funnel: analytics problems that are business problems you need to iterate and solve. Then, as we go down the stack, we have two data platforms – they're highly interoperable, designed around use cases that makes sense, and you can't go put your data anywhere else: we won’t put resources against it. So you cut off oxygen for the behaviours that you don't want. And you feed oxygen to the behaviours that you do want.

She adds: “It does take skill and leadership and technical expertise to do that, because you can't project manage your way through some of this. So we put controls around the areas that we don't want differentiation in, or to manage risk, so you can’t colour outside the boundaries. Then where there's low risk, we enable freedom, creativity, innovation.”

“[With that in mind] we have cross-functional teams that start with big questions, then start cranking out these models and data products, until we can prove that it creates value – we [work to] really create these minimal viable product (MVP) data models and data products. We start bottom-up, but then if it works, we do the scale-up bit from the top-down. A great example was a field force product called ‘Next Best Action’ where we saw benefit in one brand and decided to implement it across 20 brands and 20 markets in six weeks,” she says. ("Next Best Action" combines data and analytics to recommend to sales teams the next best steps for their interactions with healthcare professionals.)

Ramakrishnan adds: “I would say that having a data platform strategy, where data is sensibly pipelined in, ahead of what the use cases need, has been instrumental in doing that.”

Infrastructure strategy

Going back to those platforms and the broader infrastructure side of things, a key focus for her here is in ensuring that the compute fabric and analytical capabilities to make this data and AI powered vision of drug discovery a reality are world-class. Her approach, she says, is “building the road in front of the car”: “We don’t spend £100 million pounds, get a Data Platform set up and move everything in there, and somehow hope that magically it gets used” she adds. “That hasn't worked well for us…”

“We’ve taken an approach whether for data or compute that’s [based on understanding] the use cases and keeping the platform ahead of it.”

About two years ago, she explains, GSK agreed that we “don't believe we need to be in the business of running data centres. So we have exited to a multi-cloud strategy. We happened to lean into two partners who are doing really well as the AI pivot came. We're building our research data platform on Google and we will be scaling our GCP infrastructure quite a bit. We’ve also worked [directly] with NVIDIA in the past and we'll keep looking for direct partnerships in areas where we're building our own LLMs or we're training our own models; working with an appropriate AI infrastructure company as needed. But for 95% of the enterprise use cases, we'll just go to the hyperscalers and scale with them.”

(GSK’s CDTO admits that there’s work to do and the company is still one year away from getting all the existing legacy landscape onto the cloud – but emphasises that pace of cloud adoption is not a key concern: “Almost all the innovation we are pursuing is happening in the cloud and much of our application footprint including ERP will be in the cloud next year.”) 

In terms of those two data platforms “our research data platform where we put all our scientific data is called Onyx and with Google we are building it out in the cloud at scale. We have also been investing in an Azure Spark-based data platform called ‘Code Orange’ internally, which is our enterprise data platform; we've developed an interoperability strategy to make sure we can serve end-to-end use cases across both.” 

Engaging with startups

A common lament of many innovative startups is how it is to access decision makers at large multinationals. How does Ramakrishnan feel about engaging with early stage technology companies, The Stack asks?

“I'm very comfortable with engaging with the startup ecosystem” she adds, however, whilst also “recognising that we are a big regulated company that has enormous trust responsibilities to our stakeholders.”

“But for example, in the security space, we are constantly scanning what the evolving security startups are; where does it make sense to lean into something? I don't think any CIO can truthfully say today that they can't lean on the startup ecosystem for security problem-solving, although it's painful at times, but we have figured out how to do it. I think similarly, on the technology side, I am a big fan of sensible open source adoption. So we're keeping an eye on LLama and other other open source models that are evolving, particularly around the Gen AI LLM evolution, but we're looking to see how the various partners are evolving…” 

It’s a lot of spinning plates to keep up in service of GSK’s mission. Stepping back from these responsibilities and looking at the lessons from her career thus far, what would she impart to those looking to step up to similar leadership roles. She takes a moment’s pause before responding: “It's hard to drive change, if you don't have a clear understanding of the target state and what you're steering the big ship to. If you're driving a small boat, you can make a quick change. 

“But when you're steering a big ship, it's really important that you make that one degree change in the right direction: It will sail to a completely different continent if you do that at the right time, and set the right direction. The essential part of that is clarity of mission. You don't need certainty about how you get there; you do need clarity and competence. 

“The second lesson for me is around leading with empathy.

Having been in companies that really did transform themselves  dramatically, my takeaway is you can't really fix what you don't care about and you have to find people where they are, not where you are. 

“I think that’s been one of my sometimes tough learnings; when you try to go faster than a system can metabolise. Yet you have to remember that when you are pushing against the flywheel, it seems like a wall – but when it does turn and momentum builds, there's real ease and flow and that almost Sisyphean effort of pushing the rock up the mountain seems worth it when you've broken through that momentum barrier!” 

Join peers following The Stack on LinkedIn