Postdoctoral Research Fellow – Beyond Classical Graph Indices: Topology-Informed Descriptors for Molecular Property Prediction

Science Field

Produce two (2) publications as the First Author in ISI Web of Science indexed journals (Q1/Q2 in the Journal Citation Report (JCR) only).

Non-Science Field

Produce two (2) publications as the First Author in Scopus indexed journals (Q1/Q2 in SCImago Journal Rank (SJR) or JCR). Publications in JCR journals are highly encouraged.

•FPD publications/KPIs will be considered as additional KPIs to the existing KPIs of the Supervisor set by the University.

This KPI is the primary criterion for evaluation and consideration for reappointment (if applicable).

•Awarded a Doctor of Philosophy (Ph.D.) degree.

•Good academic merit and research output in the last three (3) years, particularly in publications:

Science Field

Two (2) publications in journals indexed in the ISI Web of Science (Q1/Q2 in the Journal Citation Report (JCR) only) with one (1) of the publications as the First Author.

OR One (1) Patent.

Non-Science Field

Two (2) publications in journals indexed by Scopus with one (1) of the publications as the First Author. Preference is given to publications in JCR.

•The field of appointment is open but must be relevant to the supervisor identified according to the school/center of excellence.

About the Project: We are inviting applications for a postdoctoral research position in an interdisciplinary project at the intersection of graph theory, topological data analysis (TDA), and machine learning. The project explores whether topology-informed descriptors, particularly those inspired by persistent homology, can offer richer structural representations of molecular graphs than classical graph indices such as Wiener, Zagreb, and related descriptors. The broader aim is to develop new mathematical and computational tools for molecular property prediction, with applications in QSPR and QSAR studies. Role of the Postdoctoral Fellow: The postdoctoral fellow will be involved in both the theoretical and computational aspects of the project. This includes developing topology-informed descriptors for graph-structured data, implementing them computationally, evaluating them through machine learning pipelines, and comparing them with existing graph-based descriptors used in molecular studies. Why this project matters: Classical graph descriptors have played an important role in molecular modelling, but they often summarize structure through only a limited set of handcrafted quantities. This project asks whether ideas from topology can reveal deeper and more informative structural patterns in graph-based molecular data. The successful candidate will contribute to research that connects rigorous mathematics with modern data-driven modelling. Deliverables: The successful candidate will contribute to the mathematical development, computational implementation, and machine learning evaluation of new descriptors for graph-structured molecular data. The project is expected to produce high-quality research outputs, including reproducible computational workflows, conference presentations, and TWO ISI Web of Science (Q1/Q2) manuscripts submitted to reputable indexed journals.