Water transport through carbon nanochannels can exhibit exceptionally low friction, yet experiments reveal strong dependences on nanotube radius, confinement and electronic properties. A promising explanation is quantum nanoscale friction, where charge fluctuations in water exchange momentum with electronic excitations in the carbon surface, possibly driven by its dynamical response. This Master project will extend this framework from graphene to carbon nanotubes and realistic nanoporous carbons. Using molecular dynamics, the student will compare non-polarizable, metallic constant-potential and Drude-polarizable descriptions of the carbon wall to disentangle the effects of roughness, curvature, confinement and electronic response, establishing how confinement modifies the coupling between water dynamics and the carbon substrate. The approach will then be transferred to realistic templated nanoporous carbons with controlled pore size and disorder, designed in our group in a previous thesis. Friction and permeability will be related to structural descriptors such as pore size, curvature and defect density, as well as to the electronic response of the carbon framework. Machine-learning analysis of the simulation database will identify the descriptors that best predict nanofluidic transport, providing a route toward AI-assisted design of carbon materials with optimized water-transport properties.