Internet of things (IoT) technology is booming, and from the viewpoint of embedded experimental teaching, it is worth thinking about how to lead students to conduct preliminary exploration in the field of Internet of things with embedded technology. In order to reduce the difficulties and costs of student learning and development, we designed and implemented a scalable IoT teaching and development system that integrates management and development functions. We adopted the open-source framework of Spring Boot and Vue to complete the development for the cloud side of the system. Using a micro-service architecture, we solved the typical challenges of strong coupling and poor scalability. In order to cope with high concurrency scenarios, we designed a load balancing optimization algorithm based on threshold filtering. Experimental results showed that the algorithm enhances the average response speed of the cloud and improves load balancing in complex network environments. Based on the MSP432 development platform and the EMW3080 Wi-Fi module, we implemented an easy-to-use SDK(Software Development Kit) that supports network distribution and data communication on the hardware side. The SDK reduced the complexity of the underlying hardware work and learning costs, allowing developers to focus on implementing business logic. Combining the work of software and hardware, the proposed IoT teaching development system provides a complete set of personalized IoT development solutions and an embedded teaching management system.