The share of the world’s population living in cities is rapidly increasing, and it is expected to rise to 80% by 2050. It is therefore crucial to develop new efficient and reliable methods to model the urban microclimate; in fact, these models can support urban planners and policymakers to create more comfortable and sustainable cities. High computational requirements limit existing numerical methodologies, and DANTE fits in this context and aims to create a new paradigm for fast and reliable numerical simulations bridging the fields of model order reduction, machine learning, and data assimilation. The idea is to create a research team to answer many unresolved questions in model order reduction for complex and real-life urban microclimate simulations. Particular emphasis will be given to advanced machine learning tools, which incorporate physics knowledge, aiming to improve the accuracy, interpretability, and reliability of predictive models. The identified tasks cover a wide range of different topics: dimensionality reduction of the solution manifold in problems governed by complex physical principles, uncertainty quantification, data assimilation, and inverse modeling. The new tools will have the agility of data-driven methods in complex nonlinear settings and the physical rigor of projection-based methods with quantified errors. The developed methods will significantly impact digital transformation, enabling digital twins of urban environments. Possible applications include, but are not limited to, urban air pollution, heat island modeling, wind loads on buildings, and inverse modeling approaches.