PhD in Physics-Informed Generative AI for Synthetic Energy Data in Radboud University – Netherlands

Summary

Radboud University Nijmegen is offering an NWO-funded PhD position in Physics-Informed Generative AI for Synthetic Energy Data. The project, part of the SHARE initiative, aims to develop AI models that generate realistic, privacy-preserving synthetic energy data for grid planning and decision-making, working with real-world data from Alliander. The successful candidate will design and compare deep generative approaches, embed physical constraints into data generation, build benchmark datasets, and contribute to an open-source synthetic data toolbox. This is a full-time, 4-year temporary contract, starting preferably on January 1, 2027. The salary begins at €3,204 gross per month, increasing to €4,051 in the fourth year, plus benefits. Applicants should have an MSc degree in a relevant field, a solid background in machine learning, and good programming skills. The application deadline is October 25, 2026.

Key Facts

Position Type	PhD Position
Discipline	Computer science
Research Area	Modelling tools, Programming, Computer engineering, Electrical engineering, Physics-Informed Generative AI, Synthetic Energy Data, Deep Generative Models, VAEs GANs Diffusion Models Flow Matching Gaussian Processes, Energy Systems, Power-flow, Kirchhoff laws, Graph Neural Network Architectures, Differential Privacy
Degree Required	MSc degree
Experience	Solid background in machine learning, experience with deep generative models (VAEs, GANs, diffusion models, probabilistic modelling) is a strong plus, good programming skills (Python), experience with deep learning framework PyTorch
Salary	Starting salary €3,204 gross per month, increasing to €4,051 in the fourth year, plus 8% holiday allowance and 8.3% end-of-year bonus.
Funding	NWO-funded SHARE project
Contract Duration	4-year contract, with an initial 1.5-year temporary contract followed by an evaluation, and if positive, an extension for 2.5 years.
Deadline	2026-10-25
Employer Address	Houtlaan 4, 6525XZ Nijmegen, Netherlands

About the Project

The NWO-funded SHARE project aims to develop physics-informed, domain-constrained generative models for energy-system data. The core scientific contribution of Work Package 3 involves designing and comparing deep generative approaches (VAEs, GANs, diffusion models/flow matching, Gaussian processes) for realistic load, generation, and voltage time series. The project will embed physical constraints, such as power-flow consistency (Kirchhoff's laws), operational bounds, and network topology through graph neural network architectures, into the data generation process. It also involves building validated benchmark datasets and an evaluation framework to assess statistical fidelity, temporal and spatial structure, physical plausibility, and downstream task performance. Collaboration with other researchers on integrating differential privacy into the generative pipeline is also a key aspect. The ultimate goal is to contribute to an open-source synthetic data toolbox for use by DSOs, municipalities, and researchers in the Netherlands.

Key Responsibilities

• Design and compare deep generative approaches (VAEs, GANs, diffusion models/flow matching, Gaussian processes) for realistic load, generation, and voltage time series.
• Embed physical constraints into data generation, including power-flow consistency (Kirchhoff's laws), operational bounds, and network topology through graph neural network architectures.
• Build validated benchmark datasets and an evaluation framework covering statistical fidelity, temporal and spatial structure, physical plausibility, and downstream task performance.
• Collaborate with a fellow PhD candidate and a postdoctoral researcher on integrating differential privacy into the generative pipeline.
• Contribute to an open-source synthetic data toolbox for DSOs, municipalities, and researchers.
• Publish at top machine learning venues.
• Spend up to 10% of time on teaching activities, such as assisting in courses of computing science programmes.

Required Skills

• MSc degree in computer science, artificial intelligence, data science, applied mathematics, physics, electrical engineering, or a related field.
• Solid background in machine learning.
• Experience with deep generative models (VAEs, GANs, diffusion models, probabilistic modelling) is a strong plus.
• Good programming skills, specifically Python.
• Experience with the deep learning framework PyTorch.
• Enjoy interdisciplinary work and interacting with privacy researchers, legal scholars, and energy-sector practitioners.
• Good command of spoken and written English.

Who Should Apply

This PhD position is suitable for candidates with an MSc degree in computer science, artificial intelligence, data science, applied mathematics, physics, electrical engineering, or a related field. Applicants should have a solid background in machine learning, with experience in deep generative models (VAEs, GANs, diffusion models, probabilistic modelling) being a strong advantage. Good programming skills, particularly in Python and with the PyTorch deep learning framework, are required. The role involves interdisciplinary work, so candidates who enjoy interacting with privacy researchers, legal scholars, and energy-sector practitioners are encouraged to apply. A good command of spoken and written English is essential. Prior knowledge of energy systems is not required.

Benefits

• Starting salary of €3,204 gross per month, increasing to €4,051 in the fourth year.
• 8% holiday allowance and 8.3% end-of-year bonus.
• Option to choose between 30 or 41 days of annual leave with full-time employment.
• Good employment practices, including primary and secondary employment conditions.
• Arrangements for work-life balance, flexible working hours, various leave arrangements, and working from home.
• Ability to compose part of employment conditions, such as exchanging income for extra leave days or reimbursement for a sports subscription.
• Good pension plan.
• Room for responsibility to develop talents and realize ambitions.
• Various training and development schemes.

About the Employer

Radboud University Nijmegen is the organization offering this PhD position. The project is part of the NWO-funded SHARE project and involves collaboration with Alliander, working with real-world operational data. The university emphasizes good employment practices, offering flexible working conditions and opportunities for professional development.

Frequently Asked Questions

What is the main goal of this PhD project?

The main goal is to develop physics-informed, domain-constrained generative AI models for energy-system data to create realistic, privacy-preserving synthetic energy data for grid planning and decision-making.

What is the duration of the contract?

The contract is for 4 years, starting with a temporary 1.5-year contract, which will be extended by 2.5 years upon positive evaluation.

What is the starting salary for this position?

The starting salary is €3,204 gross per month, which will increase to €4,051 in the fourth year.

Are there any teaching responsibilities?

Yes, the PhD candidate will be expected to spend a small part of their time (up to 10%) on teaching activities, such as assisting in courses of the computing science programmes.

Is prior knowledge of energy systems required?

No, prior knowledge of energy systems is not required, as consortium partners will provide domain context.

What are the benefits regarding annual leave?

With full-time employment, the candidate can choose between 30 or 41 days of annual leave instead of the statutory 20 days.