Urban air pollution rate has grown to alarming state across the India. Most of the cities are facing issue of poor air quality which fails to meet standards of air for good health. It is indeed necessary to develop an air pollution measurement and prediction system for a smart city. This proposed work acquires carbon dioxide and carbon monoxide level in the air along with Global Positioning System (GPS) location by using pollution detection sensor and uploads into Azure cloud services. Low cost embedded Beagle bone board along with gas sensors are used for data acquisition. Microsoft's Azure Machine learning service is used to predict the pollution metrics with the help of previous data. Processed data is fetched and represented by Power BI tool. Calibrated gas sensor data is fetched from sensors and successfully uploaded into cloud. Data stored in cloud is utilized by different cloud services to make the data meaningful. Proposed system is implemented and useful to monitor and reduce the pollution in a smart city by avoiding the pollution causes. © 2017 IEEE.