The cooling-slope (CS) casting technique is one of the simple semi-solid processing (SSP) processes a foundryman uses to produce the feedstock. This study attempts to develop mathematical regression models and optimise the CS parameters process for predicting optimal feedstock performance, which utilises tensile strength and impact strength to reduce the number of experimental runs and material wastage. This study considers several parameters, including pouring temperature, pouring distance, and slanting angles for producing quality feedstock. Hence, multi-objective optimisation (MOO) techniques using computational approaches utilised alongside the caster while deciding to design are applied to help produce faster and more accurate output. The experiment was performed based on the full factorial design (FFD). Then, mathematical regression models were developed from the data obtained and implemented as an objective function equation in the MOO optimisation process. In this study, MOO named multi-objective Jaya (MOJaya) was improved in terms of hybrid MOJaya and inertia weight with archive K-Nearest Neighbor (MOiJaya-aKNN) algorithm. The proposed algorithm was improved in terms of the search process and archive selection to achieve a better feedstock performance through the CS. The study’s findings showed that the values of tensile and impact strengths from MOiJaya_aKNN are close to the experiment values. The results show that the hybrid MOJaya has improved the prediction of feedstock using optimal CS parameters.