Multiple PhD Positions in Symmetries and Their Breaking in Deep Learning University of Amsterdam in Netherlands

AMLab at the University of Amsterdam is recruiting multiple PhD students for a research project on symmetries and their breaking in deep learning, supervised by Dr. Nabil Iqbal and supported by a Simons Foundation grant.

This PhD opportunity sits at the intersection of machine learning, physics, and mathematics. The project asks how deep neural networks and modern architectures can be understood using general theoretical principles such as symmetry, universality, and ideas from statistical physics. You will investigate the dynamics of deep learning from both a mathematical and empirical perspective, with the goal of developing new deep learning architectures that are more robust, efficient, and controllable.

Research topics mentioned in the advertisement include symmetry breaking in deep learning, coherent information propagation through many layers, and generative models such as diffusion models and autoregressive text generation. The project also explores whether principles from non-reciprocal thermodynamics and statistical physics can help explain generalization and guide the design of new models.

As a PhD researcher, you will conduct independent theoretical work, develop novel algorithms, run empirical studies, train and evaluate deep learning models on a high-performance computing cluster, and present results at international conferences, workshops, and journals. The position is embedded in a new group within AMLab, where you will collaborate with other students and postdocs working on deep learning from a physical point of view.

The appointment is for 4 years and comes with a full-time contract (38 hours per week). The gross monthly salary ranges from €3,204 to €4,051, excluding holiday allowance and year-end allowance. The position should lead to a PhD thesis, and the university will provide an educational plan including coursework and international meetings. Teaching support is also part of the role, including labs, tutorials, and supervision of Bachelor’s and Master’s students.

Applicants should have an MSc in artificial intelligence, computer science, physics, mathematics, or a related field. Strong mathematical skills, familiarity with machine learning and transformers, and programming ability in Python or related languages are expected. Experience with NumPy/SciPy, PyTorch, or JAX is preferred. English fluency is required. The start date is flexible, but no later than Fall 2027.

Apply online through the application portal before 2026-12-08. Your application should include a detailed CV with months for education and work experience, a motivation letter, publication list, complete Bachelor’s and Master’s course records with grades and grading-system explanation, and contact details for two referees.