Funded PhD - hypergraph machine learning for chemistry

CoReACTER; Dublin; On-site; EUR 23,000 - 28,000; Full time

Benefits offered by this job

Stipend €25,000 per year

Fees covered

Teaching in undergraduate laboratories

Job summary

The School of Chemistry at University College Dublin invites applications for one doctoral (PhD) position starting January 1, 2026, to join the CoReACTER group. You will work on computational methods for ML on chemical reaction networks, including directed hypergraph neural networks, with applications to electrolyte decomposition and biomass pyrolysis.

The successful candidate will have a Bachelor's (or Master’s) in chemistry, chemical engineering, or a related field, strong programming skills,

Qualifications

• Bachelor's degree in chemistry, chemical engineering, materials science or related field.
• Experience in software development/engineering in at least one general-purpose language (Python, Julia, C, C++, Go, Fortran, Rust).
• Experience in data science and/or machine learning is desirable.
• Familiarity with atomistic simulation techniques (DFT, MD) is preferred.

Responsibilities

• Devising and implementing ML algorithms on hypergraphs/CRNs and contributing to open-source codebases.
• Constructing CRN datasets from literature and simulations for training/testing.
• Designing CRN ML architectures for chemical property and reaction link prediction.
• Predicting outcomes of complex reactive processes via atomistic simulations and CRN ML.

Skills

Programming languages

Machine learning

Teamwork

Scientific communication

Education

Bachelor's degree

Master's degree (preferred)

Tools

Python

Julia

GPU programming

Job description

One doctoral (PhD) position (4 years) is available to work in the Community of Researchers Assessing Chemical Transformations and Exploring Reactivity (CoReACTER) within the School of Chemistry at University College Dublin. The doctoral student will be supervised by Dr. Evan Spotte-Smith, an Assistant Professor in the School of Chemistry. This position is intended to begin January 1, 2026.

The aim of this project is to develop and apply computational methods for statistical learning on chemical reaction network (CRNs) structures. Of particular interests are (directed) hypergraph neural networks, which can be used to predict vertex (molecule), hyperedge (reaction), or hypergraph (network) features. While this project will be heavily focused on algorithm design and software development, especially in its early stages, we also intend to employ the methods developed for such applications as electrolyte decomposition in metal-ion batteries or biomass pyrolysis.

One scholarship of stipend €25,000 per annum is available for 4 years for the successful applicant. Fees will be also covered. There will be a requirement to teach in undergraduate laboratories and tutorials as part of the scholarship.

Tasks:

1. Devising and implementing algorithms for machine learning (ML) on hypergraphs generally and CRNs specifically and contributing to related open-source codebases
2. Constructing and disseminating CRN datasets from the prior literature and from atomistic simulations to be used in training, testing, and evaluating CRN ML models
3. Designing and developing CRN ML architectures for tasks of chemical interest, including reaction property prediction and reaction link prediction.
4. Predicting the properties and outcomes of complex reactive processes (e.g., electrolyte decomposition, biomass pyrolysis) via a combination of atomistic simulations and CRN ML

Desired Qualifications:

• Hold a Bachelor's (required) or Master's (preferred) degree in chemistry, chemical engineering, materials science and engineering, or a related field;
• Emphasize teamwork, collaboration, and building positive community;
• Have experience in software development/engineering in at least one general-purpose programming language (e.g., Python, Julia, C, C++, Go, Fortran, Rust). Experience in Julia is a plus, as is experience with scientific, numerical, and/or GPU programming;
• Have some prior experience with data science and/or machine learning;
• Hold values such as honesty, modesty, collectivism, justice, kindness, and curiosity and are committed to the ethical practice of science;
• Are familiar with atomistic simulation techniques, including but not limited to density functional theory and molecular dynamics;
• Have skills in (scientific) written and oral communication and want to apply these skills to share research findings in scientific meetings/conferences and journals; and
• (Preferably) have some exposure to (hyper)graph theory, network science, and/or reaction mechanism/CRN studies.

Candidates who do not meet all of these criteria should not feel discouraged.