Diabetes, a chronic medical condition characterised by elevated levels of glucose in the blood, creates difficulties in regulating glucose balance, leading to potential severe health implications. The management of diabetes has been revolutionised by continuous glucose monitoring (CGM) devices; however, their usefulness is limited because they are not usually integrated with data on physical activity. To address this gap, this study describes a cloud-based remote patient monitoring system that seamlessly connects continuous glucose monitoring (CGM) devices with physical activity tracking to improve diabetes management. Using machine learning techniques on cloud platforms such as Google Cloud Platform (GCP), the system makes real-time adjustments to insulin dosage based on estimates of blood glucose levels. The cloud-based architecture makes data processing and storage scalable, safe, and efficient, making it possible to manage massive volumes of continuous monitoring data effectively. CPU utilisation peaked at just 10% during server testing with a load of 10 users per compute engine instance, suggesting strong performance and scalability. Post-testing, CPU utilisation remained low at 0.63\%. Typical usage situations produced 1.03 KiB/s outgoing and 0.53 KiB/s inbound data rates and write data rate was 6.39 KB/s despite constant read rates. The use of physical activity data improves the prediction accuracy of glucose levels, enabling customised and prompt responses. Through an intuitive mobile application interface, the system provides patients and healthcare providers with real-time data retrieval and remote access. Preliminary testing demonstrates robust performance and positive feedback, paving the way for further refinements, integration optimizations, and clinical trials to evaluate real-world effectiveness. Keywords: Continuous glucose monitoring (CGM), Physical activity, Google Cloud Platform(GCP), Machine learning. ALL ARE CORDIALLY INVITED