We focus on some perspectives about Reduced Order Methods (ROM) for parametric Partial Differential Equations (PDEs) with a special interest in parametric problems arising in offline-online Computational Fluid Dynamics (CFD) with a special attention to the enhancement provided by Scientific Machine Learning (SML). Efficient parametrisations (random inputs, geometry, physics) are very important to be able to properly address an offline-online decoupling of the computational procedures and to allow competitive computational performances. Current ROM developments in CFD include: (i) a better use of stable high fidelity methods, to enhance the quality of the reduced model too, also in presence of bifurcations and loss of uniqueness of the solution itself, (ii) capability to incorporate turbulence models and to increase the Reynolds number; (iii) more efficient sampling techniques to reduce the number of the basis functions, retained as snapshots, as well as the dimension of online systems; (iv) the improvements of the certification of accuracy, established on residual based error bounds, and of the stability factors, as well as (v) the guarantee of the stability of the approximation with proper space enrichments. All the previous aspects are quite relevant -- and often challenging -- in CFD problems to focus on real time simulations for complex parametric industrial, environmental and biomedical flow problems, or even in a control flow setting with data assimilation and uncertainty quantification. Some model problems will be illustrated by focusing on few benchmark study cases, for example on simple fluid-structure interaction problems and on shape optimisation, applied to some industrial and environmental problems of interest.