Writer identification is a popular and active research area with numerous applications such as in banking, criminal justice, access control, and detecting the validity of handwritten letters. In this research paper we present a lot of things about designing of an automatic hand writer recognition system for Amharic language by using Convolutional Neural Network and support vector machine. The handwriting styles of an individuals varied from time to time; the reason behind these types of writing styles invariant in each individual are mood, time, space, speed of writing, writing medium and tool. Even though; if it is difficult to identify these writing styles; designing an automatic hand writer system by using Convolutional Neural Network and support vector machine with the integration of applying different features like shape, orientation and angles of character possible to identify the writer of a given document. Writer identification system it is very mandatory especially; In following areas such as forensic expert decision-making systems, network security, digital rights management, biometric authentication in information and document analysis systems, writer identification is critical. According to this study, the process of writer identification, has three basic parts: preprocessing, feature extraction, and identification or classification of the document to the relevant class. In the pre-processing phase; The process of excluding extraneous information in the input data that can negatively affect recognition has been carried out by image resizing, noise reduction, and threshold segmentation. Feature extraction: is a type of dimensionality reduction where a large number of pixels of the image are efficiently represented in such a way that interesting parts of the image are captured effectively. From the total dataset of 2100, 80% (1680) used for training, and 10% (210) used for validation, and 10% (210) also used for testing the model. The testing accuracy of the model is 96%. We used accuracy, precision, recall, f1-score, and confusion matrix to evaluate our model. The model is a combination of CNN-Gabor with SVM Classifier. Key-Words: SVM, Convolutional Neural networks, Handwritter recognition