Job Description

Your responsibilities

CERN has recently launched the Efficient Particle Accelerators (EPA) project to better exploit its accelerator complex in terms of reliability, efficiency, and beam performance. Alongside classical automation concepts, machine learning (ML) techniques have started playing a major role in realizing these goals.

You will join the Data Science for Beam Operation (DSB) section within the Controls Software and Services (CSS) group. The DSB team is responsible for designing and implementing numerical optimization and ML algorithms to tackle previously intractable problems in machine operation. This includes delivering accurate, fast-executing online models and developing operational ML tools for CERN’s accelerator complex, alongside evolving the underlying frameworks.

As part of an interdisciplinary team with expertise in accelerator physics and operation, ML and optimal control, as well as computer science, you will:

• Collaborate with domain experts to design, develop, test, and deploy operational solutions aimed at automating accelerator operation;
• Conducting research to develop new algorithms;
• Develop and implement on-demand and continuous controllers to automate specific operational procedures of the CERN injector complex from start to end;
• Collaborate closely with systems, operations, and domain experts to devise robust and reliable solutions;
• Plan and carry out machine development sessions to develop, evaluate, and ensure the performance and robustness of the controllers on the accelerator and under operational conditions;
• Conduct research towards new control algorithms;

• Contribute to the evolution of existing in-house optimization and automation frameworks;
• Present findings internally as well as at international workshops or conferences and stay up-to-date with the latest developments in ML and AI.

Your profile

Skills and/or knowledge

• Strong foundation and hands-on experience with ML methods and optimal control algorithms;
• Experience in applying ML to real-world problems and in making solutions operational;

• Experience with Bayesian optimization, reinforcement learning, and/or model predictive control;
• Advanced programming skills, preferably in Python, and familiarity with relevant ML libraries, such as PyTorch;
• Strong problem-solving skills and the ability to work independently;
• Excellent communication and collaboration skills, with the capacity to work in an international and multidisciplinary team.

Advantageous skills:

• Knowledge of accelerator physics;
• Research experience in the development of control algorithms.

Language requirements:

• Fluent in English, the ability to work in French would be an advantage.