For the first time, measurements of the q-profile have been made within the first second after plasma initiation in the JET Advanced Tokamak (AT) regime. The data, obtained at both 2.0T and 2.7T using the Motional Stark Effect (MSE) system, show that the q-profile is shear reversed inside (r/a) ~0.6 with a region of very low current density in the plasma centre. This demonstrates that deep shear reversal is generated in large volume JET plasmas by an optimised plasma initiation and early current ramp phase without the need for non-inductive current drive. The magnetic shear is more negative and the negative shear region is larger at higher magnetic field when the same current waveform is used, possibly linked with the observed stronger n = 1 MHD activity at the edge at the lower values of q-cylindrical seen at 2.0T. Interpretative simulations of the current ramp phase have been performed with the TRANSP code to test the sensitivity of the modelled current profile evolution to the initial q-profile shape assumed. Compared to an assumed initial condition with weak magnetic shear, simulations starting with deep shear reversal, as measured, can retain a significant difference in q0 (~15%) as qmin reaches 3 (a typical starting point for main heating in JET AT experiments). By the time qmin reaches 2 the effect of the initial q-profile is no longer significant, indicating that modelling of plasmas in the hybrid regime, where main heating is typically applied when qmin approaches unity, is less sensitive to the initial q-profile assumption. Previous modelling of the current ramp phase of JET AT experiments, assuming a broad initial current density profile and neoclassical resistivity, produced current profiles that were too peaked compared with MSE measurements when qmin reached 1.5. A similar discrepancy is apparent when modelling more recent experiments, despite the use of a realistic initial q-profile from the measurements described above. Also, analysis of the first 1.5 seconds of a plasma discharge with early MSE measurements shows that the modelled current penetration into the plasma core is again too rapid compared with the measurements, even if, to test the sensitivity of the modelling to measurement uncertainties, Zeff was arbitrarily set to unity. These results contrast with the agreement of modelling and MSE data for stationary conditions and suggest the inconsistency may be due to factors such as measurement uncertainties or inaccuracies in the modelling. Inconsistencies between measurement and modelling in the highly dynamic current ramp phase have been observed on other devices and this issue should be addressed for validation of predictive simulations for present and future machines.