Master Thesis Autonomous Agentic Control Design - Translating System-Theoretic Properties into Self-Improving Synthesis Loops

• Your thesis explores how autonomous meta-agents can automate this end-to-end design process for complex controllers. The core objective is to translate modern control theory into structured, automated steps. Rather than relying on trial-and-error code generation, the agent must analyze critical requirements, such as constraint satisfaction, state limitations and safety boundaries, to make methodologically grounded decisions.
• You will design a reasoning framework that will guide the agent in selecting and synthesizing advanced control architectures. This framework should balance nonlinear methods (e.g., exact linearization, backstepping and sliding mode) with optimization-based approaches, such as model predictive control and reinforcement learning.
• Practically, you will build a closed-loop simulation framework where the agent interacts with physical benchmarks using Python-based control and learning libraries. The agent will iteratively write control code, run simulations and programmatically verify safety and performance before deployment.
• This project offers a unique opportunity to work at the direct intersection of generative AI and safety-critical control systems, developing a verifiable system that combines AI adaptability with rigorous engineering guarantees.

Do you have experience in Spark?, Do you have a Master's degree?, * Education: Master studies in the field of Cybernetics, Engineering, Mathematics, Computer Science or comparable

• Experience and Knowledge: profound knowledge of control engineering and machine learning; experience with Python and agentic AI
• Personality and Working Practice: you are a person with an autonomous, systematic working practice and analytic thinking
• Work Routine: partially mobile working is possible
• Languages: very good in English

At Bosch, we shape the future by inventing high-quality technologies and services that spark enthusiasm and enrich people's lives. Our promise to our associates is rock-solid: we grow together, we enjoy our work, and we inspire each other. Join in and feel the difference., Control engineering is transitioning. While traditional design relies on manual selection and calibration, agentic coding and self-improving software loops enable a new paradigm. The primary engineering bottleneck is shifting from developing a specific control algorithm to orchestrating an autonomous system that can discover, synthesize and tune the optimal controller for any plant.