Cardiovascular diseases, such as Arrhythmia and Congestive Heart Failure, are the leading cause of global mortality among the working population. While electrocardiogram remains the gold standard for assessment, early prediction is crucial. A ResNet-inspired deep 1D-Convolutional Neural Network is designed for the detection and classification of cardiac arrhythmias and other cardiac disorders from electrocardiogram signals. The residual blocks with skip connection employed in our framework adopt a unique architecture based on a “Compress–Process–Expand” strategy for hierarchical extraction of ECG features while capturing both local and global temporal dependencies in ECG data. The proposed framework was rigorously evaluated using well-known datasets, including the MIT-BIH Arrhythmia, BIDMC-Congestive Heart Failure, and MIT-BIH Normal Sinus Rhythm databases. Our approach successfully classified different types of arrhythmias in accordance with the AAMI guidelines, achieving an accuracy of 99.99%, a sensitivity of 99.76%, and an F1-score of 0.9986. Additionally, it was able to distinguish Arrhythmia and Congestive Heart Failure from Normal Sinus Rhythm, with an accuracy of 99.86% and a sensitivity of 99.80%, respectively, surpassing several recent state-of-the-art methods.