Machine Learning has gotten attention in the healthcare industry for the competences to ameliorate disease prediction. Machine learning has already been used in the health sector. Diabetes can also trigger the permanent loss of kidney function. Diabetic kidney disease (DKD) is one of the most recurrent diabetic micro vascular issues and has become the dominant cause of chronic kidney disease (CKD).Therefore, there is a need for a machine learning model and application that can effectively predict and track the level of diabetes along with Diabetic kidney disease. In this paper, our key motive is to find an efficient machine learning model to predict diabetes and diabetic kidney disease (DKD). Since, Disease Prognosis is a sensitive issue, it is not ethical to provide a result without extensive testing. Therefore, we have assessed our model using Recall, F-1 Score, Precision, AUC and also followed some robust evaluation metrics such as ROC, Sensitivity and Specificity to appraise performance of the models from the medical perspective. We are able to obtain an optimized prediction models using LightGBM with an accuracy of 98.75% on diabetic kidney disease prediction and CatBoost with accuracy of 96.15% on diabetes prediction. We have also proposed a web application using our prognostic machine learning model to predict the result based on user input. This application can be used to predict the initial stage of the diabetes mellitus and diabetic kidney disease which may help to expedite the existing disease medication process.