Electric vehicles (EVs) have had challenges with battery pack failures common in recent years caused by short circuits, thermal imbalances, fires, or even explosions. This paper explores a machine learning-based approach to the safe operating envelope (SOE) of lithium-ion batteries, aiming to address these challenges. The primary objectives of the study are to identify the key factors that impact battery safety and to establish safe charging and discharging protocols using real-time data. Describing the conditions and the criteria that guarantee battery work in EVs is basically what the determination of the SOE involves. The focus of this work is on one of the key elements of the SOE, where machine learning models consider variables such as voltage, temperature, current, and state of charge (SOC). To increase accuracy, the study integrates the creation of numerical data with data-driven modelling, which enables the prediction of ranges in which the battery can operate safely with considerable accuracy.