Transformation in Trials
A podcast about the transformations in clinical trial. As life science companies are pressured to deliver novel drugs faster, data, processes, applications, roles and change itself is changing. We speak to people in the industry that experience these transformations up close and make sense of how the pressure can become a catalyst for transformation.
Transformation in Trials
Democratizing Drug Development Through AI-Native Platforms with Ilya Burkov
Cloud computing is transforming biotech by offering purpose-built infrastructure that supports AI-driven drug discovery and development while meeting strict regulatory requirements. Dr. Ilya Burkov explains how Nebius provides full-stack solutions that democratize access to powerful technology, enabling researchers to achieve breakthroughs that previously required generations.
• Cloud computing market was built for general purpose workloads but biotech needs specialized infrastructure for sensitive data and AI models
• GPUs enable parallel processing that accelerates AI applications—like "a whole classroom solving math problems at once" versus CPUs solving one at a time
• Applications include drug discovery, genomics, protein structure modeling, quantum chemistry, and single-cell modeling for cancer treatment
• Nebius provides full-stack solutions with hardware and software layers, working with NVIDIA to offer specialized packages
• Democratizing access to AI infrastructure is leveling the playing field between small biotechs and large pharmaceutical companies
• Scientists can now accomplish in their lifetime what previously would have taken multiple generations of researchers
• Breaking down silos between data teams and institutions is crucial for accelerating healthcare innovation
________
Reach out to Ivanna Rosendal
Join the conversation on our LinkedIn page
Welcome to another episode of Transformation in Trials. I'm your host, ivana Rosendegg. In this podcast, we explore how clinical trials are currently transforming so we can identify trends that can be further accelerated. We want to ensure that no patient has to wait for treatment and we get drugs to them as quickly as possible. Welcome to another episode of Transformation in Trials. Today, we're going to be focusing on the topic of AI-powered cloud infrastructure for biotechs, and with me in the studio to discuss this topic, I have Dr Ilya Burkov. Welcome, ilya.
Speaker 2:Thank you for having me, Ivana. Thank you.
Speaker 1:Can you tell me a little bit more about what you're currently up to and why this is a topic of interest to you?
Speaker 2:Great, yeah, great to be here. So my name is Dr Iurkov. I'm the Global Head of Healthcare and Life Science Growth here at Nebius. I'm based in the UK. I've been leading the vertical for about 11 months now. I joined with about 15 years of experience in healthcare and life science sector. Before Nebius, I was working in AWS cloud services, so at Amazon, which was a really great experience and my first insight into cloud compute and the magic that it entails. Prior to that, I was involved in several different healthcare startups across the UK and Europe and early in my career I worked in orthopedics at Brooks Hospital in Cambridge where I worked on biomarker identification of early disease onsets Bridge, where I worked on biomarker identification of early disease onsets, things like osteoarthritis, osteoporosis, looked at regenerative medicine and dabbled in some biomedical engineering as well.
Speaker 1:Very cool, Good broad experience.
Speaker 2:Yeah.
Speaker 1:Well, just to set the stage for the topic that we'll be diving into, yeah, can you tell us more about what is the market for cloud computing now, and why do we need a different solution for biotech?
Speaker 2:Absolutely so. The cloud computing market is massive and it's still growing, but I think what people don't realize, or not many people realize, is that most of it was built for general purpose workloads, things like web apps, e-commerce or enterprise IT. Biotech and healthcare and life science sector, however, operates very differently. These teams are dealing with huge sensitive data sets things like omics, data imaging data, patient information and increasingly, they're training large AI models that require not just raw compute, not just the hardware, but the very specialized infrastructure and software, and that is the regulatory. So, to add to that regulatory security, reproducibility demands that are very unique to life sciences. You don't get a lot of this in e-commerce or web apps and so on, and it's clear that one size fits all does not work when it comes to this. Biotechs need infrastructure that's purpose-built, but it's optimized for scientific workflows, compliant by design, but it also adds the ability to support AI from discovery all the way to clinical development, and that's the gap that we at Nebius are addressing.
Speaker 1:Very interesting. What does compliance by design look like when we're talking about cloud computing?
Speaker 2:Yeah, it's from the start when you talk about how the data is stored, where the data is stored, looking at data sovereignty and everything that comes with it. Different geographies have different regulations, have different compliance requirements. Making sure that that is all taken care of is key, and making sure that the infrastructure itself handles the data in a safe and secure manner.
Speaker 1:I see, and if we go like a step deeper into the infrastructure here, if you were to explain to perhaps a high school student, how would you explain what a GPU is and why is it important?
Speaker 2:I get that a lot, actually, Even explaining to my family or other people, friends that might not be working in this industry. Imagine your brain doing maths. A CPU is like a really smart person solving one hard math problem at a time. That's a central processing unit. A GPU, which is the graphics processing unit, is like a whole classroom of students involved all together into solving lots of math problems all at once. So it's that accumulation of computational power that entails. So originally GPUs were built to handle graphics in video games, like drawing thousands of pixels very, very quickly on a screen. But it turns out that the same ability to do tons of small calculations in parallel makes them perfect for AI, where models that are designed by software engineers and others they need to process huge amounts of data at lightning speed. So if you're training an AI to recognize cancer cells or design new medicines, a GPU would let you do that work faster than any regular computer chip that you would find like a CPU.
Speaker 1:Now, that's very cool and a nice and simple explanation. So if we Consider the use that you just described for this sort of cloud computing, it sounds like you're talking about using it for drug discovery, basically as one of the use cases. I'm curious are there any other cool ways that your solution has been used so far?
Speaker 2:Yeah, so absolutely, we cover a plethora of different solutions, but I would say that we focus on drug discovery, drug development, genomics, multiomics. There is an aspect of biotech, health tech, medtech as well, so looking at imaging data analysis and so on. We've seen a really wide range of high impact solutions built on our platform, everything from AI models that predict drug target interactions to kind of fully automated pipelines for analyzing genomic and imaging data. So some biotech teams are training foundation models on protein structures or small molecules. Others are using our infrastructure to power various types of labs and running real-time data learning loops. We work with companies that are hired by pharma companies to integrate some of their workflows. We work with groups that are looking at single cell modeling, where they are using a virtualized approach to say which drug is more effective than the other for oncological treatments targeting tumor cells. We are working with groups that are using it for quantum chemistry. I think in terms of possibilities, it's endless within this and it's not just in one particular segment of this healthcare and life science vertical.
Speaker 1:Very cool. And how much of the infrastructure do you provide and do people build applications on top? Or how much of the stack do you offer?
Speaker 2:So, yeah, we are a full stack solution. The approach that we take is that we give access to the GPUs, but not just the hardware. We have the layer of software on top to handle things like managed Kubernetes, managed Slurm. We have some proprietary work as well. We work very closely with NVIDIA to offer a lot of their NIMS packages, things like Bionemo and Parabricks. It comes with a plethora of other models that are very useful for this and, together with all of these parts, we're handling the pre-training data preparation side. We can handle the training side and the inference side all within one platform. This is quite unique and makes us kind of stand out from the other Neo clouds and the hyperscalers that are out there. We're kind of very comfortably sitting in the middle. What ties all of these things together is basically secure a high-performance AI native infrastructure specifically designed for this.
Speaker 1:Very cool. So if we talk about more the compliance part, which is always interesting for my listenership, how do you ensure that this is specifically well-suited for health and safety information that we often store?
Speaker 2:Yeah, sure. So I mean we are working very closely with non-patient identifiable information. At the moment, the European regulations are different to that of the US. With the US, there's a need for things like HIPAA, and that's a separate topic. We also look in GDPR. In the EU, there's the EU AI Act, there's data residency requirements and so on. When you are talking about the different data sets that we work with at the moment, we're focusing on drug discovery and drug development. They don't need patient identifiable information. We cover that primarily. At the moment, we are looking at finalizing some of the other compliances that are required to handle patient identifiable information, and we're going to be starting that in the next few months.
Speaker 1:Very cool. Also exciting to take that next chunk. Has there been a request for that from your customers?
Speaker 2:Absolutely. Yeah, I mean, we just got started about 11 months ago within this vertical, started about 11 months ago within this vertical and since then we've been exponentially growing and getting all of the things that are necessary in line. So this has definitely been on our roadmap and we're very excited to be doing this in the next few months very cool.
Speaker 1:Um, so we talked a little bit about the different kinds of regulation and that you must adapt to. One of the things that I find interesting about your company is that you are european based. Do you feel like it makes a difference in the world as it is today to be based out of europe?
Speaker 2:yeah, I mean we are european based, but we do have a global footprint and we're globally focused um as um. We had a quick discussion previously. We are a NASDAQ listed company. We have a global footprint. We're building one of the world's leading AI centric platforms.
Speaker 2:For me, the work that we've been doing in Europe in particular, has been very, very significant. Being Europe based does make a difference to the local communities and the sovereign AI approach we have in Europe. We have data centers in France, in Finland, in Iceland and there's a brand new one coming in the UK very shortly, which we're very excited about. But we also have data centers in Kansas City, missouri. We have a data center coming in New Jersey as well. So we're not excluding the other markets.
Speaker 2:Current geopolitical climate we try not to focus on that. We focus on the engineering side and what we do best. For me, it's the healthcare and life science side, and many of our partners are looking to have infrastructure that's high performing. They have regional routes. Sometimes they want to have data remaining in Europe. Other teams are working across the world and they might want to have some data in Europe, some data in the US. So we adhere to what the customer needs and, yes, it gives us a clear advantage around Europe and we work very closely with NVIDIA in Europe. They use us as a preferred partner and when we're looking at how we can work around the regulatory landscape in Europe, we are involved in a number of different projects that are European-based.
Speaker 1:Okay, very good. I am curious from your perspective how did you end up setting on that? This is the problem that you want to solve in healthcare so in general in general, like there are so many things one could start working on but solving like the actual platforms that we work on, why was that's the most interesting thing to do?
Speaker 2:so as a, as a company, this was a direction we wanted to take. For me personally, I think that the platform helps to unlock a world of new possibilities where drug development becomes not just faster but fundamentally more personalized and inclusive, predictive. For me, I imagine a future where a researcher can design, test and simulate a new therapy in days, not years. I want to make sure that they're using real world data, generative AI and automated labs running all within our infrastructure, where clinical trials can adopt to patient outcomes in real time, where rare diseases can't be ignored because of the costs involved to develop the treatment and the costs are too high. I'd love to see Nebius play a key role in making that more democratic. Access to AI-driven science.
Speaker 2:At the moment, there's a lot of vendor lock with companies and not providing access to the right GPUs, not providing access to the right infrastructure. Small biotech or academic lab does not always have the same tools in their arsenal. We want to break that habit. We want to enable everybody to have the freedom of choosing their tools and their chips the ones that they want to use, no matter the size of their company. Chips, the ones that they want to use no matter the size of their company. So we work with startups that are 5-10 people in numbers who are getting the same GPUs and the same technology that huge pharma company are getting, and we want to make sure that they all have visibility to that, basically, leveling the playing field and delivering better healthcare not just faster but fairer and, in the end, it's about making the infrastructure invisible and the impact at the heart of it and making the impact visible in a form that leads to real therapies reaching real patients, and as soon as possible.
Speaker 1:Very cool, and if you were to imagine what would be the coolest thing that could happen on your platform, what would that be?
Speaker 2:the coolest thing, yes, okay, uh, I would say that probably connecting the cutting-edge technology to the complexity of the human biology. Early on, I worked in computational biology and various other things and and I was fascinated by how data can can help us understand diseases in a systematic level. That led me to kind of understanding what we are working on. If we can enable as many startups, as many enterprise-based companies, as many clinical environments with this infrastructure underneath and can scale so that they can collaborate across the teams to make these breakthroughs in innovation, that's key.
Speaker 2:And one example I like to give is I had a discussion with a scientist at a university who said typically, it would take them three PhD students and around four years per person to understand the interaction of a molecule, and the work that they were doing has always been thought of how much can I achieve in my lifetime before I have to hand over to the next generation to continue the work? And he said that with the power of generative AI and with the power of AI GPU use, he has been able to accelerate that work into achieving what he wants to get within his lifetime and not have to limit himself to studying everything at a one-by-one status. He can do so many things sequentially and have his PhD students working on things one after the other rather than just focusing all the time on one thing. That has now become a reality, and I want to see more of that, not just on an academic level, but also on an industrial level and a personalized medicine level.
Speaker 1:That is very cool, this whole acceleration of basically everything, of creation, of discovery, that we can achieve when AI is by our side.
Speaker 2:Absolutely.
Speaker 1:Very cool. Well, as we start rounding off, I always ask, I guess, the same question in the end, and that is if I was to give you the transformation trials magic wand that has the ability to change one thing in the life sciences industry, what would you wish to change?
Speaker 2:That's a great question, yeah, and I would say that if I could change one thing about the industry, I would say that I would break it down in the silos between data teams and institutions.
Speaker 2:So much more innovation gets delayed, not because the science isn't ready, but because the systems aren't connected. Whether it's between kind of discovery and clinical, or's between discovery and clinical, or between pharma and regulators, or even between academic labs and industry, valuable insights get trapped. So if we can make collaboration as seamless as computation, I believe we'd accelerate not just the speed of drug development and discovery, but also the quality and equity of what we deliver to patients. The future of healthcare depends on our ability to think and build beyond the walls that we've inherited, and having the right infrastructure without the vendor locking in place. Having access to the latest and greatest things that we provide in NVIDIA, together with NVIDIA sorry, at Nebius is key. I would summarize it as that that's a great answer.
Speaker 1:Well, if my listeners would like to ask you further questions or they just want to learn more about Nebius, where can they go?
Speaker 2:I would encourage them to join us on the website nebiuscom. Reach out to me either through LinkedIn or vice versa. Please just have a look at the incredible stuff we're doing. Look at the case studies that we can again share with your listeners, some of the work that we've done in the customer stories that back up everything that I've said.
Speaker 1:Thank you so much, that's awesome.
Speaker 2:Thank you very much for your time, Ivana. Thank you so much for tuning in to another episode of Transformation in Tri.
Speaker 1:Everyone, thank you so much for tuning in to another episode of Transformation and Trials. If you have suggestions for a guest for a show, then reach out to me, ivana Rofendale, via LinkedIn. If you'd like to support the show, you can now buy us a coffee. Find a link in the show notes. Thank you so much for tuning.
.png)