Research Fellow in AI and Machine Learning for Materials Discovery (Alchemy) Grade 7

A key part of the discovery workflow is the efficient identification of stable chemical entities from the vast space of possibilities. This requires the ability to predict structure from composition. While there are many ways demonstrated to do this, simple inversion of a chemical composition to a low-energy structure based on data is not possible today. This project will use machine learning and symbolic AI in combination with exact optimisation methods (VV. Gusev, et al., ‘Nature’, 2023, 619, 68- 72) to produce the most efficient extended structure prediction algorithms known. In this context, we are relying on computer science to improve existing practical crystal structure prediction tools and to develop radically new approaches to this problem. At the centre, we are leveraging combinatorial optimisation techniques such as local search and integer programming as well as different first and second order continuous optimisation methods. We are seeking an exceptional candidate with skills is one or more of the following areas:

• Artificial intelligence
• Machine learning
• Combinatorial optimisation
• Reinforcement learning
• Verification
• Quantum computing

Your application should demonstrate your experience and the relevance of your skills to the project. The project team combines computer scientists with experts in crystal structure and materials synthesis. We have already used human-in-the-loop decision support to realise outperforming functional materials, specifically solid lithium electrolytes (G. Han, et al., *Science*, 2024, 383, 739-745). While the structure prediction tools used there were efficient, this and our other work reveals the urgent need for fast and reliable structure prediction – the advent of machine-learnt potentials makes the use of such tools to evaluate compositions of direct relevance to experiment feasible, further increasing the value of the new tools we will develop in this project. In order to extend the scope for human input, we aim to extend the capabilities of machine learning methods with symbolic reasoning approaches that incorporate expert insight. This tool will be a key component in the human-in-the-loop workflow being developed by the AI for Chemistry Hub, and you will have the opportunity to work with the teams unique (B. Burger, et al., *Nature*, 2020, 583, 237-241) experimental robotics tools and capabilities. This will accelerate the application of the new structure prediction tools to realise materials in the laboratory. This broad activity rests on a unique combination of theory and practice: theoretical results about crystals and their symmetries are used to improve practical tools and algorithms that lead to discovery and characterisation of new materials.

We are looking to recruit a Research Fellow to work on one of the forerunner projects of AlChemy, namely “Human in the Loop”, which aims at integrating cutting edge AI technologies to accelerate the discovery and synthesis of new materials. This part of the project led by Prof. Matt Rosseinsky OBE FRS.

Commitment to Diversity

The University of Liverpool is committed to enhancing workforce diversity. We actively seek to attract, develop, and retain colleagues with diverse backgrounds and perspectives. We welcome applications from all genders/gender identities, Black, Asian, or Minority Ethnic backgrounds, individuals living with a disability, and members of the LGBTQIA+ community.

For full details and to apply online, please visit: recruit.liverpool.ac.uk

Advert information