As countries seek to decarbonize their energy systems, green hydrogen has emerged as a promising energy carrier. This paper studies the production of green hydrogen from offshore wind in the Dutch North Sea, with particular emphasis on understanding how system design decisions and uncertain parameters affect key performance indicators. The analysis builds on a detailed techno-economic simulation modeling framework that is used to accurately simulate energy flows and estimate the levelized cost of hydrogen. To address the strong dependence of simulation outputs on input choices, and the computational infeasibility of evaluating the full input space, this study applies “optimization with constraint learning”, where supervised machine learning models are trained using simulation data. This approach allows one to approximate the relevant input-output relationships and replace the computationally expensive simulation framework with surrogate machine learning models that are more tractable and can be embedded in mixed-integer optimization problem formulations. The methodology provides a fast and flexible decision-support tool, offering both descriptive insights and prescriptive guidance for decision-making. The results the effectiveness of such an approach for analyzing trade-offs and exploring stakeholder-defined objectives and constraints in a dynamic and computationally efficient manner.