Pillar VC has announced that applications are now open for the Encode AI for Science Fellowship powered by the Advanced Research + Invention Agency (ARIA). The Fellowship will place top AI talent across seven leading UK science labs working in ARIA’s opportunity spaces, which cover climate, science, robotics, neuro-technologies and more. Each AI fellow selected for the one-year Encode Fellowship will receive a competitive stipend, access to state-of-the-art computing resources, and the opportunity to work on pressing scientific challenges.
Pillar VC has identified 16 projects for Encode AI fellows, and two of these projects are being led by DPAG scientists.
Rui Ponte Costa will lead the project on AI for Systems-Behavioural Neuroscience
He says, ‘We are developing an AI-powered technology that can infer the contributions of specific brain regions to behaviour by analysing gameplay data, enabling non-invasive diagnosis and treatment of neurological disorders. Our approach leverages state-of-the-art deep reinforcement learning models, specifically MuZero and Dreamer architectures, which learn to solve complex tasks in ways that mirror human cognitive processes.’
The system will be trained and validated using a combination of Atari game benchmarks, simulated behavioural data, and real-world functional MRI recordings from collaborators at Harvard University and the University of Oxford (in DPAG). This technology aims to revolutionise neurological treatment by enabling personalised, non-invasive therapies that target specific brain regions, potentially transforming how we diagnose and treat a broad spectrum of neurological conditions from behavioural data alone.
Dame Molly Stevens will lead a flagship project on AI for Biomaterials Design
She says, ‘We are building an automated platform that uses artificial intelligence to design and optimise nanoparticles for medical applications. Our system combines robotic laboratory equipment with AI algorithms that learn from experimental results to rapidly discover optimal nanoparticle recipes, replacing the current time-consuming trial-and-error approach. The platform will process hundreds of different nanoparticle formulations simultaneously, using real-time measurements of particle properties to guide the AI in selecting the most promising combinations. By automating and accelerating nanoparticle development, we aim to reduce optimization time from months to days while creating better-performing particles for medical applications, potentially accelerating the development of new treatments and therapies.’
Read detailed descriptions of all 16 projects (and how to apply) here