<p>This study employed machine learning techniques to design a nickel-based superalloy with both high creep life and low coefficient of thermal expansion (CTE). Initially, a database containing creep life and CTE values was constructed, and the mutual information (MI) algorithm was then used to calculate the MI values between the features and creep life. The feature variables with strong correlations were screened out based on the MI values. At the same time, a three-step feature selection method identified the key feature descriptors of CTE. Then, based on the screened features, the gradient boosting regression (GBR) algorithm was applied to construct two performance prediction models. After training, the coefficients of determination (<i>R</i>²) of the models on the test set reached 0.88 and 0.85, respectively. Finally, the two GBR prediction models were optimized using a genetic algorithm. Following single-objective optimization, the creep life increased by 12.18%, and the CTE was reduced by 10.02%. In multi-objective optimization, the designed Ni_62.5Ta₁₁Cr₈Al_5.5W₅Re₄Mo₃Co₁ alloy exhibited a creep life of 6784.08&#xa0;h and a CTE of 12.50<i> e</i>⁻⁶&#xa0;°C⁻¹. Compared to the original database, the creep life improved by 12.13%, and the CTE decreased by 6.92%. In addition, the shapley additive explanations (SHAP) method was introduced to improve model interpretability. This research provided a theoretical foundation for the efficient design of nickel-based superalloys.</p>