Handwritten hindi character recognition using convolutional neural network. Accuracy of 96.14% achieved which is better than the earlier works done in this field. It used two basic network architectures, Lenet5 and GoogLenet. It also involved a web application that was developed in python using Flask.