Doctoral Research Associate; Wissenschaftliche*r Mitarbeiter*in, level E TV-L

42,500 students and 7,750 employees in teaching, research and administration, all working together to shape perspectives for the future – that is the University of Münster. Embedded in the vibrant atmosphere of Münster with its high standard of living, the University’s diverse research profile and attractive study programmes draw students and researchers throughout Germany and from around the world.

The Institute of Organic Chemistry in the Faculty of Chemistry and Pharmacy at the University of Münster is seeking to fill the position of a PhD student in the Faculty of Chemistry. The position is available from the earliest possible date, preferably by 1st April 2026, as a full (100 %) position limited to 3 years. This PhD position is part of the EU‑funded Marie Skłodowska‑Curie Doctoral Network on Low Data Machine Learning for Sustainable Chemical Sciences under Grant Agreement No. .

As part of the Glorius Group at the University of Münster, we are a team of passionate researchers committed to shaping the future of molecular science. Our research bridges catalysis, functional molecule design, molecular machine learning and data‑driven discovery to address pressing scientific challenges. We are pioneers in the application of data science to challenges in (organic) chemistry, with over 15 years of interdisciplinary research.

Our data science team develops smart screening strategies and machine learning tools to accelerate reaction discovery & analysis, thereby improving chemical understanding. With a strong publication record and a collaborative spirit, we offer an inspiring environment for PhD candidates aiming to grow scientifically. We love science, innovation, and shaping the future of digital chemistry – together with you.

This project is a collaboration between 13 academic and industrial organisations with 14 PhD students in total. The aim of LowDataML is to train a new generation of scientists at the interface of machine learning, chemistry and other fields. We propose a data‑science guided and ML‑driven screening and optimisation workflow to improve the generality of modern synthetic methods.

Our first objective is to identify structural patterns and scaffolds that are accessible by modern synthetic methods but are at the same time underrepresented in databases of bioactive molecules such as ChEMBL. To achieve this, we will perform direct substructure matching and 3D similarity searches to determine key reactions based on product motifs. Generality of a methodology is critical to allow the incorporation of a variety of substrates and functional groups.

We will use an additive‑based screening approach to rapidly evaluate the generality of key reactions, followed by reoptimisation of promising reactions using a Bayesian optimisation algorithm. Batch selection strategies will allow parallel experimentation in 96‑well plates, with analysis by liquid and gas chromatography combined with UV, mass and flame ionisation detection. Optimised reactions will be applied in multi‑step diversity‑oriented synthesis of drug‑like compounds.

• Identification of underrepresented structural motifs accessible by modern synthetic methodologies.
• Assessment of the robustness of key reactions using an additive‑based screening approach.
• Development of a multi‑objective Bayesian optimisation algorithm for the reoptimisation of reactions.
• Application of the reoptimised key reaction in the synthesis of drug‑like compounds.

As a PhD student in this Doctoral network you will use machine learning (ML) for the optimisation of the generality of synthetic methods. You will start with the search for structural patterns and scaffolds that are accessible by modern synthetic methods but are at the same time underrepresented in databases of bioactive molecules. You will apply an additive‑based screening approach to identify key reactions which can be investigated further.

The applications of ML as a tool for the optimisation of reaction yields will be developed and implemented. Optimised reactions will be run in parallel using a liquid handler to set and work up experiments in 96‑well plates. Analysis…