The University of Luxembourg, a leading international research university, is offering a fully funded PhD position in the field of uncertainty-aware machine learning force fields (MLFFs) within the Faculty of Science, Technology and Medicine (FSTM). The position is part of the FNR-funded UMLFF project, which aims to advance atomistic simulations by developing next-generation MLFFs with built-in uncertainty predictions. This research is crucial for enabling safe, automated active learning and creating broad, reliable MLFFs that deliver quantum-chemical accuracy for large systems.

As a doctoral researcher, you will join the Theoretical Chemical Physics group led by Professor Alexandre Tkatchenko and supervised by Dr. Igor Poltavskyi. The project is highly multidisciplinary, combining physics, chemistry, materials science, and machine learning. Your research will focus on uncertainty estimation in deep neural networks for MLFFs, implementing and testing uncertainty-aware loss functions, studying calibration and post-calibration for predictive uncertainty, and integrating uncertainty modules into MLFF architectures. You will also analyze MLFF feature spaces to detect out-of-distribution atomic environments, use chemical neighborhood representations to identify sparse or unseen cases, and combine ML features with chemical descriptors to enhance uncertainty robustness.

Further tasks include building uncertainty-aware general purpose models, merging developments into unified MLFF architectures, training MLFFs on large, diverse datasets, and evaluating models through molecular dynamics simulations and benchmarks. You will integrate uncertainty-aware MLFFs into active learning frameworks, explore automated dataset generation for molecules and materials, and contribute to open-source tools for automated MLFF training.

Applicants must have a Master's degree in Physics, Chemistry, Materials Science, Computer Science, or related fields, with a strong background in theoretical or computational physics or chemistry. Experience with machine learning, Python, and modern ML frameworks (PyTorch/JAX) is required. Desired skills include familiarity with atomistic simulation codes (ASE, FHI-aims, VASP, CP2K), knowledge of deep learning architectures, graph neural networks, or uncertainty quantification, and experience with HPC environments. B2-level proficiency in the language of the thesis is mandatory.

The University of Luxembourg offers a modern, dynamic, and international environment with high-quality equipment and close ties to the business world and Luxembourg labour market. The yearly gross salary for the PhD position is EUR 41976 (full time). The contract is for 36 months, full-time, based at the Limpertsberg Campus. The university promotes an inclusive culture and encourages applications from individuals of all backgrounds.

To apply, prepare a single PDF file including your CV, cover letter, transcript of university-level courses, and contact information for two references. Applications must be submitted online through the HR system; email applications will not be considered. Early application is encouraged as applications are processed upon reception.