About the Project
The School of Engineering Mathematics invites applications for a PhD at the intersection of Artificial Intelligence, mathematics, and molecular science. This is a unique opportunity to advance your expertise in graph modelling, geometric deep learning, and Graph Neural Networks (GNNs) to drive next-generation AI tools for molecular discovery, including novel drugs, advanced materials, and sustainable technologies.
Research directions include (but are not limited to):
• Novel GNN architectures for handling structural and geometrical invariances/equivariances (e.g., 3D molecular symmetries) to enable enhanced generalization and data efficiency in low-data regimes such as few-shot learning.
• Expressivity beyond standard message-passing, lightweight graph transformers capable of capturing higher-order interactions at scale.
• Generative modeling on graphs (e.g., diffusion-based generative models) for de novo molecular design with integrated uncertainty quantification.
• Multimodal conditional modeling, incorporating diverse experimental constraints (e.g., embedded spectroscopic signatures, reaction conditions) to guide graphs/molecule generation and optimization.
• Efficient ML techniques including pruning, compression, model fusion, to enable scalable and resource-efficient graph learning and inference.
This project focuses on core algorithmic and mathematical development in graph-based AI. Collaboration with the School of Chemistry and Prof. Butts’s lab, renowned for expertise in machine learning, computational chemistry and spectroscopy, and closely collaborating with pharmaceutical industry, provides opportunities to validate models against real-world experimental data and contribute to impactful projects in drug discovery and materials engineering.