The air pollution is one of the key factors in the threat posed by the environment and results to so many health complications. Today’s air quality control is not able to bring changes to the current conditions focusing on the decision rules. Therefore, in this paper, a Smart Air Quality Control System (SAQCS) is proposed which is a combination of CNN-LSTM and Transformer based structures with Autoencoder models for forecasting the air quality and for managing the Indoor Air Quality (IAQ). Concerning the live sensor measurements, PM2.5, CO2, VOCs, temperature and humidity are used by the system to control the ventilation standard and the operations of the purification system. There are shortcomings prevailing in standard approaches because they do not involve changing environmental adjustments. As proved by the experimental results, the developed system has reduced pollutants by 55%, improved energy efficiency by 30% and reduced system response time by 20%. This can be deemed as the research that has been done as the evaluation of intelligent and dynamic management of air quality in indoor environment. To address the issue of having static and rule base method of controlling pollution and attaining a scalable morality of regulation, SAQCS use the deep learning models. According to the research finding of the study, there is notable performance enhancement in the air quality management and attainment of heathier and sustainable indoor environment through deep learning ensemble models.