PhD in Geometric Deep Learning for Materials Science

About the role

The Heidelberg Institute for Theoretical Studies (HITS) invites applications for a fully funded PhD position in Geometric Deep Learning. The research focuses on developing generative models for crystalline materials, supporting future design of batteries and carbon capture catalysts.

Key responsibilities

• Conduct cutting‑edge research on diffusion models, flow matching and other generative approaches for materials science.
• Integrate concepts from differential geometry and group theory to exploit crystal symmetry in model design.
• Publish results at top machine‑learning conferences (NeurIPS, ICML, ICLR) and high‑impact journals.
• Collaborate with the SIMPLAIX consortium (University of Heidelberg, KIT, HITS) and supervise interns and master’s students.

Required profile

• Master’s degree in computer science, mathematics, theoretical physics or a related field.
• Strong theoretical foundation in machine learning and solid linear‑algebra knowledge.
• High motivation, ability to work in an international team and good communication skills.

Required skills

• Programming proficiency (e.g., Python)
• Linear algebra
• Differential geometry (optional but beneficial)
• Representation theory (optional)
• Mathematical optimization
• Experience with generative models, diffusion models, flow matching

What we offer

• First‑class research environment with outstanding computing infrastructure.
• English‑speaking, international work atmosphere.
• Strong connections to academic and industrial labs.
• Competitive salary between 55,000 EUR and 60,000 EUR per year.