Handwriting recognition is a significant application of pattern recognition. As mobile technology expands, handwriting recognition software needs to be developed, since today the majority of students rely on digital machines such as tablets for their studies, so the proposal of a system to recognize the most common vocabulary in computer science can be a good choice. In this paper, a deep learning-based Convolutional Neural Network (CNN) is employed to perform handwritten recognition using a newly produced dataset with a total of 27682 handwritten word images for 500 different classes, The structure's model comprises13 layers, 4 convolution layers, 4 -subsampling layers, and a fully connected component that have 2 hidden layer and an output layer in addition to batch normalization and dropout layer, for the purpose of the recognition challenge, the three optimizers Adam, RMSprop and SGD with momentum are used, the best validation accuracy attained is 97%, by using SGD with momentum