<p>The rapid evolution of malware obfuscation techniques, such as code packing, encryption, and metamorphic transformations, has significantly reduced the effectiveness of single-model detection systems. These advanced evasion techniques complicate the identification of obfuscated malware, creating a critical need for more robust and adaptive detection approaches. This research aims to design and evaluate an effective ensemble learning framework capable of accurately detecting obfuscated malware. To achieve this goal, we propose three ensemble configurations, combining multiple machine learning models. These models are optimized using hyperparameter tuning techniques to improve accuracy and reduce overfitting. This optimization process enables each classifier to operate under its most effective configuration while maximizing the complementary strengths of the ensemble. The proposed models are evaluated on obfuscated malware dataset, that has a total of 58,596 records, evenly split between 29,298 benign and 29,298 malicious samples. Experimental results demonstrate that the proposed configurations outperform existing methods, achieving higher detection accuracy in all the scenarios evaluated. The findings confirm that ensemble learning is a powerful approach to address the challenges posed by obfuscated malware.</p>