Silviu Craciunas

DTU Compute is offering PhD scholarships in machine learning for the optimization of cyber-physical systems . The project focuses on developing learning-based methods that can speed up analysis and improve the optimization of systems where exact solutions are too costly to compute in practice. The selected PhD student will work in the Embedded Systems Engineering section at the Technical University of Denmark and join a research environment centered on embedded, distributed, and cyber-physical systems. The research direction combines predictive modelling with optimization: building surrogate models for expensive analytical calculations and simulations, then using AI-driven methods to search for good configurations and decisions under practical computing constraints. Research problems mentioned in the advert include predicting quantities such as execution delays, energy use, and resource demand; guiding configuration, scheduling, and resource allocation; and assessing solution quality, uncertainty, transferability, and agreement with original models and constraints. Application contexts include software-defined vehicles, datacenters, industrial automation, and energy systems, giving the project a broad but clearly systems-oriented character. The PhD topics will be defined together with the supervisors. The position is embedded in a department with a large PhD community and an international research culture, and students are expected to publish, collaborate with project partners, and contribute to teaching and student supervision. The advert also highlights the broader DTU Compute environment, which combines research in mathematics, data science, computer science, and computer engineering with work on AI, machine learning, IoT, cybersecurity, and related digital technologies. Supervision is provided by Prof. Paul Pop and Associate Prof. Silviu Craciunas .