Diabetes is a medical condition that has affected millions around the world and still doing it at an increasing rate. Many researches show that an early detection of diabetes can prevent risk-factors that can be caused as a result of this disease. The inclusion of machine learning and deep learning algorithms in the early prediction of diabetes has played a big role in the health-care monitoring system. Many of the early researches put emphasis on improving the accuracy of prediction models but often, the datasets available are too short for deep learning algorithms to leverage their true potential. In this research, along with a highly accurate deep learning model, a new system has been proposed integrating cloud services where users can directly contribute to enrich an existing dataset which also can be used in improving accuracy of the deep learning methods. The model proposed in this research shows promising results on predicting diabetes and the proposed system ensures usage from anywhere in the world with contribution in further enriching an existing dataset.