The proposed article is devoted to the study of the architecture of an information system for monitoring data from heat energy meters using Internet of Things (IoT) technologies. The implementation of remote monitoring systems for thermal energy consumption is currently one of the priority directions in energy management aimed at increasing resource efficiency and reducing operational costs. In the face of rising fuel and energy prices, as well as the global necessity to reduce greenhouse gas emissions, heat supply enterprises and facility managers require intelligent tools to control energy usage. The integration of Internet of Things (IoT) technologies allows for the automation of data collection processes, timely detection of accidents, and optimization of heating supply schedules. Therefore, the development of reliable, secure, and scalable information systems for monitoring acts as a critical task for the modernization of municipal infrastructure and the transition to "smart city" concepts. Despite the widespread use of electronic heat meters, a significant problem remains the lack of unified centralized solutions for real-time data collection and analysis. Existing systems often face challenges such as the incompatibility of proprietary protocols from different meter manufacturers, high operational costs of traditional GSM/GPRS communications, and insufficient scalability. Furthermore, ensuring the security of confidential consumer data and processing large volumes of historical information for predictive analytics requires more advanced architectural approaches than those currently available on the market. The objective of this article is to justify and design the architecture of an information system for monitoring data received from heat energy meters based on IoT technology. The system aims to ensure unified data collection from heterogeneous devices, reliable transmission with minimal latency, secure processing, and storage of large data volumes, as well as providing tools for visualization and forecasting of heat consumption. The study proposes a multi-level system architecture consisting of a Device Layer, Gateway Layer, Network Layer, and Application Layer. At the device level, the integration of commercial heat meters with microcontrollers (STM32, ESP32) and additional sensors (DS18B20) is proposed. To ensure energy efficiency and wide coverage, a hybrid communication approach is utilized: LoRaWAN technology is used for long-range transmission (up to 10 km) with low power consumption, while NB-IoT is employed for deep signal penetration in dense urban environments. The system utilizes the MQTT protocol over TLS/DTLS for secure and lightweight data transport. For data processing, a stream processing model (using tools like Apache Kafka) is implemented to filter noise and aggregate readings in real-time. The storage architecture combines Time-series databases (InfluxDB) for high-speed writing of metrological data and NoSQL databases (MongoDB) for managing metadata. The analytical component of the system leverages machine learning algorithms, specifically ARIMA and LSTM models, to forecast heat loads with an accuracy of ±5%, and K-means clustering to detect anomalies such as leaks or unauthorized consumption. The research concludes that the proposed architecture successfully solves the problem of interoperability between different equipment manufacturers through protocol unification. The combination of LoRaWAN and NB-IoT ensures reliable coverage and extends the autonomous operation of sensor nodes up to 30 days. The use of containerization (Docker) ensures the system's scalability to serve tens of thousands of devices. The implementation of predictive analytics allows for the optimization of boiler house operations, potentially reducing energy costs by 8–10%, which confirms the practical value and economic efficiency of the developed system.