The evaluation of the seismic vulnerability of existing bridge portfolios is of great concern for transport network managers especially in earthquake-prone countries. The not-negligible number of bridges designed in the past without adequate anti-seismic requirements, the poor structural knowledge of these structures and the material degradation pose challenges in this field. This study proposes a framework for large scale seismic vulnerability assessment of multi-span reinforced concrete girder bridges considering knowledge-based uncertainties. It is based on subsequent modules that involve the input of basic knowledge data (e.g., the maximum bridge span length and pier height), the simulation of knowledge-based uncertainties, simplified seismic analysis, and fragility assessment. Different Machine Learning (ML) algorithms such as Artificial Neural Networks and Extreme Gradient Boosting are investigated to define a ML surrogate model of the non-linear response of bridge piers via pushover analysis to fast the assessment process and reduce the computational effort. A case-study section demonstrates the application of the framework in the case of cylindrical reinforced concrete piers showing promising results and suggests the feasibility of extension to other bridge-pier systems.