Transitioning back to academia
Before returning to academia, I was a senior manager at McKinsey & Company where I was advising clients on digital and analytics strategy, primarily in the pharmaceutical and healthcare sectors.
It was during this time that I realised the profound and transformational impact that artificial intelligence (AI) and machine learning (ML) would have across most aspects of our lives, and I wanted to be at the forefront of this technological revolution.
To thoroughly understand, and potentially contribute to AI research and development, transitioning back to academia became imperative. I was fortunate to be accepted to a PhD programme in the Department of Computer Science at the University of Oxford. I worked under the guidance of Professor Yarin Gal, and with the support of GSK AI and ML through an Engineering and Physical Sciences Research Council (EPSRC) ICASE award.
Transitioning from the fast-paced environment of industry to academia was both challenging and exhilarating, shifting from delivering immediate, practical solutions to taking the necessary time to explore the theoretical foundations of ML and pursue longer-term objectives.
Deep generative models for biology
My research has focused on creating deep generative models to tackle challenges in computational biology and chemistry. Additionally, I’ve developed versatile tools for uncertainty quantification in high-dimensional spaces, designed compute-efficient ML architectures, and discovered optimal sets in iterative experimental settings.
I became fascinated by protein modelling, and in particular the challenge of predicting which mutations would preserve or enhance their functionality. Be it for human proteins to predict genetic disease risk or viral proteins to help with pandemic preparedness.
Inspired by the progress of transformer architectures to model natural language, I then concentrated on training large-scale protein language models for fitness prediction and design.
From ML research to real-world impact
Towards the end of my PhD, I focused on methods for the design of novel biomolecules with desired properties. The sheer potential for real-world application is staggering – from therapeutics design, to new material synthesis, breaking down plastics and forever chemicals, efficient carbon capture, and many more.
In my current role in the Marks lab at Harvard, I focus on the development of AI methods for protein engineering, and I am enthusiastic about leveraging these new approaches to support our transition towards a healthier and more sustainable future.
Reflections on my EPSRC ICASE experience
Looking back, my advice to aspiring researchers working on applied AI research is twofold.
First, engage with domain experts early on to identify both the questions that really matter and actual real-world constraints.
This is where the EPSRC ICASE PhD programme excels, as it offers a unique combination of academic rigor and practical industry relevance. In my own experience, the in-depth understanding of critical issues in the field by GSK was a standout factor, and my journey would certainly not have been the same without the joint mentorship of Professor Yarin Gal and Dr Patrick Schwab at GSK AI and ML.
Skills such as translating complex theoretical concepts into practical solutions, effective interdisciplinary communication, and navigating the dynamics of industry-academic partnerships are ones I continually rely on in my current endeavours.
Second, be bold! With the convergence of AI and biology, the forthcoming decade promises unparalleled discoveries. There is so much left for us to understand. This is only the beginning of the journey.
Read my research on deep generative models for predicting the impact of genetic mutations in humans.
Read about my research endeavours.
Top image: Credit: DKosig, iStock, Getty Images Plus via Getty Images