Mankind engaged in mining industries faces diverse challenges because of the variety of particulate matter (PM) released, which causes pollution by inhaling that released PM pollutants, and long-term exposure to that causes severe health hazards like pneumoconiosis, silicosis, asbestosis, lung scarring, cancer, and bronchitis. This study aims to develop a live-time surveillance system for PM pollutants by integrating ML and the Internet of Things to improve worker safety and environmental health. The methodology involves the design of a live monitoring system utilizing the sensor Plantower PMS5003 to measure PM1.0, PM2.5, and PM10 in mining areas. The system consists of a low-cost microcontroller ESP32 that transmits data to a display unit consisting of cloud storage like ThingSpeak for real-time results. The collected data is combined with temperature and humidity measures of the OpenWeatherMap API and analyzed using machine learning techniques including linear regression, gradient boosting regression, decision trees, random forests, and ensembling algorithms which are stacking, voting, blending, etc. to predict dust PM levels and send alerts to mobiles using message-sending API. Among the machine learning models, which has superior accuracy with the root mean squared error and mean absolute error. The integration of IoT and ML provides a robust remedy for monitoring and predicting PM levels, enhancing safety and reducing health hazards for miners.