The electrocardiogram (ECG) offers an accessible and non-invasive assessment of human health. Chagas disease, which affects nearly 6.5 million people across Central and South America, is known to have symptoms that appear in ECGs. Using time-series machine learning techniques, critical information can be extracted from these ECGs to detect Chagas disease as opposed to serological tests. As part of the George B. Moody PhysioNet Challenge 2025, we developed a classification approach consisting of two components: (1) a multi-view representation of 12-lead ECGs; (2) ensemble classification. Our team, GAIN-ECG, developed a novel approach that combines kernel-based feature extraction through MiniRocket with classical signal features, such as Heart Rate Variability (HRV), Discrete Wavelet Transform (DWT), and Fast Fourier Transform (FFT) features, through early fusion. We then employ an ensemble framework to classify the onset of Chagas disease. Testing against a held-out subset of the public training set, our model achieved a challenge score of 0.481, AUROC of 0.880, and F1 of 0.113. On the hidden validation set, our model received a challenge score of 0.090.