Protein sequence classification is one of the most important and applied areas of research in bioinformatics. It is used in several applications, from drug discovery to evolutionary relationship analysis. There are different alignment-based algorithms that are implemented for protein sequence classification. Traditional alignment-based sequence similarity checking, particularly with large datasets or long sequences turns out to be a complicated, time-consuming, and computationally expensive process. Here, the pairwise comparisons of two sequences based on string algorithms lead to a time complexity of O(nm), where n and m are the length of the first and second protein sequences, respectively, making large-scale analysis less feasible. The proposed model suggests an alignment-free model through feature-based machine learning techniques, thus in turn reduces time complexity. The Physicochemical and structural protein properties are used to train different classifiers which includes Decision Tree, Random Forest, K-Nearest Neighbors (KNN), Light Gradient Boosting Machine (LGBM), and Convolutional Neural Networks (CNN). The best of these models is the Convolutional Neural Networks (CNN) method, which classifies protein sequences with an accuracy of approximately 91%. By adding additional features to the current system, expanding the dataset, and including natural language processing, the results of the current study can be further enhanced.