job description
apply here:
Fermentation-based manufacturing is playing an increasingly important role in the global economy, spanning from food production to pharmaceuticals and the chemical industry, and it is also a key component of the green transition. Due to the complex and heterogeneous biological mechanisms underlying fermentation, achieving optimal performance typically requires fine-grained, time-dependent control of the local environment. Ideally, this optimal control could be achieved without conducting a large number of expensive and time-consuming experiments, and without relying on complex online feedback mechanisms. However, in practice, process optimization remains costly and the results are difficult to predict, which hinders the sustainability and economic viability of fermentation-based manufacturing.
To address this challenge, you will develop and test new data-efficient, open-loop fermentation control strategy optimization methods. These methods will be built upon the latest theoretical advances in optimal control theory, reinforcement learning, and numerical methods, as well as laboratory analytics and automation techniques. The ultimate goal is to achieve automation of fermentation process optimization, thereby enhancing existing fermentation-based manufacturing technologies and making new technologies feasible.
offer requirements
Your project will center on Pontryagin’s Maximum Principle (PMP), a mathematical result originally developed for designing aircraft control trajectories. You will build upon the latest results in Riemann-Stieltjes optimal control to combine PMP with Bayesian optimization, enabling data-efficient learning. Subsequently, you will implement and validate this new method in simulated fermentation processes and compare it with the current state-of-the-art open-loop reinforcement learning approaches. The next step is experimental validation: you will conduct automated control fermentation experiments in the BRIGHT biofoundry.
The following qualifications are required:
– Solid knowledge of Pontryagin’s Maximum Principle and related open-loop dynamic system optimal control theory.
– Familiarity with Bayesian optimization.
– Experience in numerical computing, including optimal control, dynamical systems, Bayesian inference, and Bayesian optimization.
– Experience in running controlled fermentation experiments.
– Experience in collaborative software development according to modern best practices.
– Excellent academic writing and presentation skills.
You must hold a two-year Master’s degree (120 ECTS credits) or have an equivalent academic level.
offer benefits/salary
DTU is a globally recognized leading technical university in research, education, innovation, and scientific consulting. We offer an international, rewarding, and challenging working environment. We strive for academic excellence in an environment characterized by collegial respect and academic freedom, while maintaining a sense of responsibility.
Employment is based on the collective agreement with the Danish Confederation of Professional Associations. Salary is determined in consultation with the relevant union. The employment period is 3 years.