PhD Scholarships in Geometric Deep Learning for Materials at DTU Energy

The Atomic Scale Materials Modelling (ASM) section at DTU Energy is offering two PhD scholarships focused on geometric deep learning for materials. These positions are part of the AI4Materials initiative and involve developing, training, and deploying generative and reinforcement learning models for 3D graphs of materials. The research aims to explore vast inorganic chemical spaces and design synthesizable energy materials, integrating physics simulation-based workflows to evaluate stability, properties, and synthesizability-by-design. Collaboration with NTU Singapore and other international partners is a key aspect of the projects. PhD candidates will develop software frameworks, APIs, and workflows for foundational machine learning potentials (MLPs), including geometric graph manifold learning methods to analyze heterogeneous quantum mechanical datasets. The work includes implementing and benchmarking state-of-the-art graph neural network (GNN)-based MLPs, building autonomous finetuning and active learning pipelines, and designing robust computational workflows. Candidates will co-develop well-documented, shareable codebases and tools, following modern software engineering practices, and contribute to an open, inclusive, and collaborative research culture at DTU Energy. The interdisciplinary environment at DTU Energy spans physics, chemistry, materials science, and computer science, with strong ties to large research centers and international collaborators. The department focuses on functional materials, components, and systems for sustainable energy technologies, including fuel cells, electrolysis, power-to-x, batteries, and carbon capture. Research is based on expertise in electrochemistry, atomic scale and multi-physics modelling, autonomous materials discovery, materials processing, and structural analyses. Applicants should hold a two-year MSc degree (120 ECTS points or equivalent) in physics, chemistry, materials science, chemical engineering, computer science, applied mathematics, or a related field. Documented experience in computational materials modelling or quantum mechanical simulations (e.g. DFT, MD), machine learning/deep learning (especially GNNs, generative models, or foundation models), scientific software development, high-performance computing, and data infrastructure is required. Proficiency in Python and modern scientific computing tools is expected, and experience with high-throughput workflows, MLOps, and open-source development is desirable. Strong communication skills in English are essential. The PhD scholarship is subject to academic approval, and candidates will be enrolled in a general degree programme at DTU. The appointment is for 3 years, with salary and terms based on the Danish Confederation of Professional Associations' collective agreement. The starting date is 1 March 2026 or as agreed. DTU offers a dynamic, international research environment with a commitment to academic excellence, collegial respect, and academic freedom. The university supports diversity and encourages applications from all qualified candidates, regardless of background. Applications must be submitted online by 19 December 2025, including a cover letter, CV, transcripts, diplomas, publication list, and references, all in English as a single PDF file. For further information, contact Associate Professor Arghya Bhowmik ([email protected]). More details about DTU Energy and the application process can be found at www.energy.dtu.dk and the job application link.