The post doctoral position is part of the AI4NRJ project. This project aims to develop a novel form of intelligent, embedded supervision for optimizing smart energy networks. Unlike existing approaches (AI, digital twins), it will simultaneously integrate adaptability to new data, new habits and robustness by considering cause-and-effect relationships. A foundation model-based AI, trained on multiple datasets and capable of performing various tasks, will be developed to handle heterogeneous data, including complex parameters like demand fluctuations and energy losses, while predicting consumption and detecting anomalies.