The atrioventricular (AV) node plays a central role in cardiac conduction by delaying atrial-to-ventricular activation, supporting subsidiary pacemaking, and filtering rapid atrial impulses. Despite its importance, human AV-nodal electrophysiology remains difficult to study experimentally, and most available computational formulations are based on non-human data or non-nodal cell types. In this work, we developed and evaluated a human-like AV nodal model using a multiscale computational framework. Starting from the human atrial Skibsbye cell model, we optimized ionic parameters toward reported AV-nodal biomarkers, including spontaneous rate, maximum diastolic potential, action-potential amplitude, peak voltage, upstroke velocity, and repolarization duration. The resulting model, denoted hAV, was then compared against rabbit nodal, human atrial, ventricular, and Purkinje reference models using single-cell biomarkers, integrated ionic-current fluxes, and APD restitution. Finally, hAV was embedded in a three-dimensional AV conduction-axis tissue model including atrial tissue, fast and slow pathways, compact AV node, and His--Purkinje system to assess emergent conduction behavior under different pacing rates. At the cellular level, hAV reproduced key AV-nodal features, including spontaneous activity, a depolarized diastolic potential (-67.55 mV), low upstroke velocity (7.25 V/s), prolonged repolarization (APD90 249.04 ms), and a spontaneous rate of 52.10 bpm. Ionic-current flux analysis placed hAV in an intermediate regime distinct from atrial, ventricular, rabbit nodal, and Purkinje-like phenotypes, while restitution analysis showed a broader and more complex dynamic profile than the reference models. At the tissue scale, the model reproduced preferential fast-path conduction at baseline and rate-dependent 2:1 filtering during rapid atrial pacing. These results support hAV as a physiologically plausible and computationally robust basis for multiscale simulation of human AV-nodal electrophysiology.