Abstract: The use of cloud computing becomes more mature for large-scale system. To optimally utilize cloud computing resource, task scheduling algorithm needed. Task scheduling algorithm approach try to minimize completion time and optimally utilize cloud computing resource. The biggest challenge of task scheduling algorithm not only how to distribute tasks to workers fairly, but also distribute task to the right worker at the right time. On the real world, the tasks may have different resource usage. There exists task with high computational usage and a task with high memory usage. The different type of task may affect performance of task scheduling algorithm. Grouping or classification of task may increase scheduling performance [1]. In this paper, task scheduling algorithm based on task profiling is introduced. The task profiling algorithm profiles the task based on its resource need, then classify it to HIGH_CPU for high computing task, HIGH_MEMORY for high memory task or LIGHT for task which not consume much resource. Using profiled tasks, proposed system is able to choose worker more precisely. Result shown, the proposed task profiling algorithm gives better completion time compared to traditional scheduling algorithm.