Efficient post-harvest sorting of kiwiberry (Actinidia arguta) is essential for maintaining fruit quality and prolonging the viability of storage and transport. Because kiwiberry is climacteric, mixing ripe and unripe fruits may accelerate over-ripening and shorten shelf life, creating challenges for commercial distribution. This study investigates the integration of hyperspectral imaging and predictive modelling to automate the sorting of ripeness in a prototype device designed for handling kiwiberry. Hyperspectral data from 1,770 fruits were processed using transformation techniques, including Standard Normal Variate, Multiplicative Scatter Correction, Savitzky-Golay filtering, and spectral derivatives. Three regression models - Multivariate Adaptive Regression Splines, Partial Least Squares Regression, and Principal Component Regression - were evaluated to predict soluble solids content as an indicator of ripeness. Based on raw spectra, the Partial Least Squares Regression model obtained the highest accuracy, achieving a determination coefficient of 0.95, root mean squared error for prediction of 0.6778, and residual prediction deviation of 4.18 on the test set. The developed system provides a foundation for real-time, non-invasive sorting, enhancing post-harvest management, reducing waste, and advancing automated fruit sorting technologies as a practical solution for optimizing supply chain logistics in kiwiberry production.