Diabetes mellitus is a chronic metabolic disorder affecting millions worldwide, requiring continuous monitoring to prevent severe complications such as hypoglycemia and hyperglycemia. Traditional glucose monitoring techniques, like finger-prick testing, are often invasive, discontinuous, and inconvenient for patients, especially in remote or resourcelimited areas. This study focuses on addressing these limitations by developing a real-time, cloud-integrated remote glucose monitoring system. The proposed solution involves an ESP32 microcontroller that simulates blood glucose values and transmits them to the cloud platform. A mobile application, developed using ReactJS, retrieves this data and classifies glucose levels as hypo-glycemia, normoglycemia, or hyperglycemia. Firebase integration ensures secure authentication, data storage, and real-time access, while additional healthcare features-such as doctor appointment booking and medical report downloads-enhance usability. Two machine learning models, trained for pre- and post- meal scenarios, enable precise classification of glucose states. The models achieved classification accuracies of 91.3% (pre- meal) and 89.7% (postmeal), with end-to-end latency under 2 seconds. This system enhances diabetes management by enabling secure, real-time monitoring and has potential for clinical sensor integration and predictive analytics in future applications.