At Isomorphic Labs, our team of world class researchers and engineers is building and further developing the powerful new predictive and generative AI models designed to accelerate scientific discovery at digital speed on our mission to solve all disease. All Isos come to this challenge leveraging their varied background and diverse perspectives to create a unique and collaborative interdisciplinary environment to achieve our mission. Ahead of ICLR, we're excited to share the stories of two leaders on Iso’s ML team - Agnieszka Podsiadło and Michael Schaarschmidt - who have reflected on their journeys to Iso, leadership and their perspective on working at Iso.
Agnieszka Podsiadło, Head of Engineering
Q: Tell us about your journey into ML and what led you to Isomorphic Labs?
A: I have always been interested in life sciences - I thought I was going to cure cancer using technology, but access to data and funding wasn’t available at the time. I moved into industry and ran ML teams at a number of the big Silicon Valley companies, but I still felt this need to have an impact. When Iso was launched, I was so impressed by its technology, ambitious mission, talented team, and its resources enabling Iso to look beyond the next quarter and think long-term. I thought if anyone is going to crack this and redesign how drugs are made, it’s going to be Iso.
Q: What do you enjoy most about working at Isomorphic Labs?
A: First, people here are generous with their time, are passionate about sharing what they know, and are incredibly curious to learn from others! One of my earliest memories at Iso was when in my first few weeks, a colleague approached me in the kitchen, introduced themselves and said “do you want to hear about my research?”. Since then, I’ve shadowed chemists and biologists, asking them how ML can help make their work better. We often have what I call ‘silly questions hour’, where it almost feels like I’m being paid to be back in school to ultimately produce something that will be hugely impactful!
Second, I’ve really enjoyed being able to build the culture on the team that I lead. It’s important for me to build a group who are unique, but bring a complementary skill set to the table where we can value our contributions equally and work in partnership together. We’ve developed a really engaging way of working where we can share the big ideas, but execute quickly once a direction of travel has been decided.
Finally, we’re working at a time when ML is having a huge impact on the world within a number of different domains. At Iso, being on the cutting edge of technology and seeing what our AI models can do feels like magic. It’s hard work, but it’s so rewarding - but not many people can say I’m waking up, tackling diseases, and having fun doing it!
Q: If you could give one piece of advice to candidates attending ICLR, what would it be? What's one thing you wish more candidates knew about working at Iso?
A: Have as many chats as possible. For me, it’s so energising when you meet people at conferences like ICLR who are passionate about their work and making a difference in their fields.
Michael Schaarschmidt, Director of Machine Learning
Q: Tell us about your journey into ML and what led you to Isomorphic Labs?
A: I was a Senior Research Scientist at Google DeepMind working on infrastructure for training large models during the time DeepMind was developing AlphaFold 2. The work happening during that period was fascinating – in a short space of time, it really began to demonstrate where AI could be applied to solve some of the largest challenges humanity faces. Consequently, I began asking myself the specific question “How can we use compute and scaling expertise to improve science applications?” Around that time, Iso was spun out. With its ambitious mission and the team that was being built to achieve it, I felt that it was the time to go all in and bring large-scale compute to science.
Q: What do you enjoy most about working at Isomorphic Labs?
A: First, we’ve arrived at a unique moment in history where technology and compute have reached a point where problems in biology which were unable to be solved either have been solved, or are now within our reach. It’s incredibly exciting to be at Iso where we’re at the forefront of developing the models which will help transform drug discovery.
Second, I love working with colleagues who are able to dream big and innovate off the back of their imaginations. On the teams that I lead, I encourage my colleagues to ‘think about the ridiculous’, asking them to consider how they would approach a problem if they have an order of magnitude more compute resources. With the pace of technological development and our growth trajectory, it's often the case that this is the level of resources available to us at Isolabs within a year - we’re regularly ahead of the game in mapping out the next stage of innovation within our models.
Finally, because we work so closely with our biology and chemistry teams, we have an incredibly short feedback loop where we can quickly understand if our models met their needs. You’d be surprised how motivating it is to see the direct impact of our work with drug designers. When they tell us, “This finally solves the problem” - it’s just incredibly rewarding!
Q: If you could give one piece of advice to candidates attending ICLR, what would it be?
A: It’s definitely a fast-paced event! I would recommend figuring out who you want to talk to ahead of time, reach out, and don’t be shy. Also, have coffee chats and participate in workshops. My colleagues and I will be around at ICLR - if anyone is interested in hearing more about Isomorphic Labs, I’d love to speak to them!
IsoLabs will be conducting an accelerated interview process for those attending ICLR, where successful candidates will receive an offer at the conference! If you are going to ICLR and would like to interview for our open ML roles, register your interest here: bit.ly/3E3k55P
Our team will also be available for 1-on-1 coffees at ICLR! If you’d like to learn more about opportunities at Isomorphic Labs, sign up here:
.png)
.svg)
