One of the problems with reinforcement learning is its black-box aspect, which makes it difficult to ensure the reliability of the model once it has been deployed in the real world. These guarantees are essential to ensure that the model does not make decisions that could have serious consequences once deployed.

The aim of this internship is to address this issue as part of our project, and to investigate solutions that will enable us to obtain stronger guarantees than we currently do, thereby increasing confidence in our models. During the internship, the trainee will have several tasks to accomplish:

• Review the state of the art in Safe RL and explainable RL, and more specifically their application in the context of energy systems and smart grids.
• Select, implement and test the most promising approaches.
• Validate these approaches on a simulator or even on a real system.
• This subject is innovative and can lead to a PhD thesis.

• Currently enrolled in an engineering school or Master's program in the Reinforcement Learning field
• Do you have experience in Reinforcement Learning and knowledge on energy systems? Do you know Machine Learning and Deep Learning? Knowledge of Safe AI or explicable AI.
• Are you comfortable with office automation and familiar with the Office suite? You've got good coding skill (Python, RL librairies, NN librairies) and are familiar to Latex?
• Are you a self-starter, rigorous and a team player? Can you take the initiative? A professional command of English will be essential for this position.