Age and gender prediction are used extensively in the field of computer vision for surveillance. Advancement in computer vision makes this prediction even more practical and open to all, thus enables the world to come up with datasets, one of which, used in this paper, is UTKFace that has 1000 pictures of male and female actors ageing from 0 to 100. In this paper, we propose a Convolution Neural Network (CNN) with ResNet50 architecture to predict age and gender. CNN is a Neural Network (NN) algorithm that extracts the deep features from the image and specifies the desired output at the final layers. Age prediction is approximately near to the real values with a five difference in both ways. Gender prediction is accurate in all the test data presented to the model. Validating with arguments shows no change in training and validation. Our model successfully executed with approximately 80% in gender prediction and 60% in age prediction that can be furtherly advanced with pipelining with other classification models and much larger real-world datasets.