Thanks to advances in Machine Learning, right now, it feels like humanity’s understanding, development, and application of technology is growing exponentially.
With the scale and pace of change the world has witnessed in the last few decades, it’s a privilege to have played a very small part in this journey, working on a range of unexplored, complex, once seemingly impossible challenges of the new technological age.
Collaborating with a team of brilliant colleagues, I helped build Amazon’s first Canadian software engineering organisation, tackling the challenges of operating a global fulfilment network, the foundations of which are still in use today. Leading the Technical Working Group of the Pan Cancer Analysis of Whole Genomes project I helped collect and analyse the world’s cancer genome data to better understand mutational processes that underlie this disease. While serving as the CTO of SOPHiA GENETICS, we analysed over one million genomic profiles, from more than 70 countries, building a unique computational platform that allowed a more comprehensive understanding of the relationship between genetics and disease.
Still, the opportunity to join Isomorphic Labs as its first Chief Technology Officer felt incredible, even within the context of the innovation I contributed to throughout my career. As I spoke to the recruitment team and learned more about the company, it became clear that coming across a team, a mission, and a technical challenge like this happens once in a lifetime - or never.
As I think about the year that’s gone by since I joined, and everything I’ve learned in the process, the reasons that compelled me to join the team are even more exciting, motivating and fulfilling today.
Driven by our mission
Although our goal is to revolutionise the drug discovery process, this is not our ultimate mission. No matter who you are, or what you do, we have all witnessed the toll disease can take on us and our loved ones.
Many of us decided to join after asking ourselves “who are we doing this for?”. For me, it's the family members or friends who might one day rely on medicines that are yet to be discovered to improve their quality of life or even save their lives.
Many of us are similarly driven by the desire to apply our skills, expertise, and the technology that is pushing the boundaries of innovation, to a mission that has a clear positive impact beyond its own commercial viability. Put quite simply, we want to use technology to bring better healthcare to everyone in the world.
Machine Learning at the core
Advances in Machine Learning (ML) are permeating our lives, yet it feels like there is so much more to come. I liken the feeling to that of the internet in the early 2000’s. There seemed to already be a website for just about everything, but it was hard to imagine just how integral the internet would become to our lives and society in the present day.
And similarly, we already see the ubiquity of ML models in virtual assistants, recommender systems, self-driving technology, and a myriad other applications. In the past year alone there’s been an incredible surge of generative models in art, gaming, and text. Yet applications of ML, and especially deep learning, to science, appear to still be in their early days.
Scientific progress is rarely easy, owing to the complexity of biological systems and our limited ability to comprehensively interrogate them. Amazing breakthroughs are nevertheless possible if one is able to match up the right scientific question with a carefully constructed dataset, an appropriate model, and robust engineering to take it large-scale.
DeepMind’s AlphaFold2 (AF2) is a prime example of the power of deep learning, applied with mastery to a well-posed scientific problem, unlocking predicted structures of over 200 million proteins.
Here at Iso, we are devout users of AlphaFold 2, forming a strong partnership with the DeepMind Science team. But AF2 alone is not enough to solve drug discovery. What we need is a coordinated set of models that can reason comprehensively at multiple levels of chemical and biological systems - modelling how molecules work and interact with each other at the lowest level, how molecules aggregate to form organelles and entire cells along with their behaviour, how cells give rise to complex organisms, and finally how these organisms interact with their environment.
To achieve this lofty task we are placing ML at the core of our endeavour, building the world’s strongest applied ML research, data, and engineering teams and throwing the full weight of our experience, creativity and resources behind these problems.
Building an interdisciplinary approach
Though ML is our superpower, solving scientific problems at this scale really takes a village. Even the most sophisticated AI algorithms are helpless if they are not grounded in firm scientific reality, one in which intuition and expertise takes years to develop.
This is why we are building our team in a truly interdisciplinary manner, bringing together computer scientists, engineers, mathematicians, chemists, biologists, physicists and many other disciplines, to move the mountain together. This means having shared goals, a common space to physically be present in, spaces to have fun and hang out in, and interdisciplinary jams to freely share ideas and educate each other.
Importantly, our research and engineering teams have been only one piece of the talent puzzle Isomorphic Labs is solving for. Given the scale and the complexity of the challenge, we’re also thinking critically about finding the right talent across a number of professions - including operations, law, finance and HR - that will enable us to build an entire company that can be innovative, flexible, and creative enough to succeed in our mission.
A startup environment within Alphabet
As we continue to grow, we’re no doubt still developing the unique collaborative environment that will underpin Isomorphic’s work. In doing so we take a considered approach reflecting our part of a larger whole of the Alphabet ecosystem of companies.
Unlike a traditional startup, in getting off the ground, Isomorphic Labs has enjoyed the basic business structures that make a company work, provided by Alphabet from the start. This has allowed the team to devote much of their focus to the science and how we work together as a group. Thanks to this support, we can be more ambitious in our work, make mistakes, and learn in the process.
We also have access to technological resources at an unprecedented scale. You need an incredible amount of computing power to develop cutting-edge ML models, and Alphabet has been able to provide us with this, and many additional internal and external tools to aid us on our journey.
It’s exciting to imagine how much is possible as we continue to build this company! The scientific and technical problems we are solving are ambitious, meaningful, and incredibly interesting.
Reaching our goal, however, will in part depend on us exploring the various aspects of how we create an environment that will enable us to do amazing things: from how we build the most diverse team possible; to how we collaborate to make scientific breakthroughs; to how we provide opportunities for colleagues to fuel their creativity outside the office, we are resolved to figuring out how to do what’s necessary to accelerate this transformative journey. I’m looking forward to the year ahead, when we’ll be even closer to our destination.