The emergence of the Industrial Internet of Things and smart factories has introduced new challenges and opportunities in industrial operations. This research focuses on automating the front-end stages of the data life cycle within a smart industrial environment, particularly emphasizing predictive maintenance and anomaly detection. Smart factories generate vast amounts of data from various sources, such as vibration sensors, temperature and humidity monitors, energy consumption meters, and visual inspection cameras. Efficient management, analysis, and automation of this data are vital for improving machine performance, operational efficiency, and reducing unplanned downtime. Traditional machine maintenance methods are often reactive or based on fixed schedules, leading to inefficiencies, premature or delayed servicing, and significant downtime. This study proposes an automated system that integrates front-end stages of data from multiple sources to predict machine failures and detect anomalies. The system uses deep learning techniques, specifically an autoencoder model, to identify anomalies in normal machine operations, enabling timely maintenance actions. The proposed study involves continuous data collection from various sensors, data preprocessing, and feature extraction to obtain key indicators. An anomaly detection model based on deep learning identifies unusual patterns in the data, indicating potential equipment malfunctions or operational inefficiencies. Multi-modal data fusion techniques provide a comprehensive view of machine health, enhancing predictive maintenance and anomaly detection.