Department researchers awarded four ERC Starting Grants in 100% success outcome

Four researchers in the department have been awarded prestigious European Research Council (ERC) Starting Grants, marking a 100% success rate in this year’s highly competitive funding round. 

The awardees are Professor Amir Goharshady (Project: SPES – Sparsity-guided Efficient SMT-solving for Program Verification), Professor Ana Namburete (Project: WOMB2COT – Computational framework to assess brain maturation in small vulnerable newborns, from womb to cot), Professor Christian Rupprecht (Project: Volute – Visual Omniversal Learning from Universal Teachers), and Dr Jiarui Gan (Project: ASPAC – Algorithmics of Stochastic Principal-Agent Coordination). 

Their projects span research in software verification, perinatal brain imaging, large-scale computer vision, and algorithmic game theory. Together, they reflect the breadth and strength of early-career research across the department. 

Dr Gan’s project aims to establish the computational foundations of the principal-agent framework, by developing algorithmic theories and tools for orchestrating ‘multi-agent symphonies’ in next-generation agentic systems. This could power applications in areas that are being rapidly transformed by modern AI – from the sharing economy and transportation management to public policy making and collective problem-solving. These advances will help to ensure that AI agents and AI systems are not only more powerful, but also more principled, more cooperative, and better aligned with the common good. 

Professor Goharshady’s project aims to make software verification several orders of magnitude more scalable by exploiting the fact that many programs have significant underlying sparsity and tree-like structure. This can be leveraged to design more efficient verification algorithms. By applying methods from advanced mathematics and computer science, such as parameterised complexity and algebraic geometry, these faster algorithms will be capable of reliably checking very large, safety-critical software systems. 

Professor Namburete’s project will extend her pioneering foetal brain imaging methods into neonatal care, bridging a critical gap in assessing infants born prematurely or growth-restricted. Using advanced quantitative ultrasound measures, the project will provide deeper insights into early brain development. Her team will integrate artificial intelligence, computational neuroanatomy, and biophysics-inspired reconstruction to significantly enhance ultrasound image quality by reducing artefacts. This innovative method will convert routine scans into accurate, high-resolution 3D images, enabling early detection of subtle developmental vulnerabilities. WOMB2COT will leverage low-cost ultrasound devices, creating a seamless imaging pathway from pregnancy to infancy, accessible directly at the bedside. 

Professor Rupprecht’s project will tackle a critical bottleneck in the development of large computer vision models: the shortage of usable, labelled data. The largest computer vision models are now trained on huge amounts of images from across the internet. But preparing this data is costly: expert labelling is expensive, and even simple labelling tasks become unaffordable when dealing with billions of images. This shortage of usable data has become a bottleneck to progress. Professor Rupprecht’s project will tackle this problem by making data an active part of the training process itself, using generative models to help overcome the labelling barrier. 

This year, the ERC awarded €761 million in Starting Grants to 478 early-career researchers across Europe. Just over 12% of the 3,928 proposals submitted were successful, highlighting the exceptional quality of the department’s applications.