DrSebastian Espinel Rios

We focus on three core areas in the context of bioprocesses and biosystems, with the ultimate goal of integrating intracellular and extracellular degrees of freedom to enable efficient, automated, and self-driving bioproduction.

• Knowledge abstraction: Digital twin technologies, model reduction, and knowledge transfer.
• Knowledge exploitation: Methods for multilevel optimization and control, including soft sensors.
• Knowledge facilitation: Software development for bioprocess digitalization and automation.

We have experience in hybrid physics-informed modeling and model-based and learning-based optimization and control. We have developed model predictive control approaches for dynamic metabolic control; hybrid metabolic cybergenetic models combining macroscopic bioprocess knowledge with machine-learning components informed by process data or mechanistic metabolic models; and automated frameworks for model selection and uncertainty quantification. Our work also spans virtual sensing and state estimation of partially observed bioprocesses, including the inference of intracellular proteins and metabolites, as well as reinforcement learning for robust multi-setpoint, multi-trajectory, and dynamic metabolic control.

Our areas of application span industrial biotechnology (e.g., chemical production) and biopharmaceutical manufacturing (e.g., recombinant protein production).