The development of IoT environment has proved beneficial in attracting new users to the platform, which has in turn led to the development of cloud computing. Efficient task scheduling has an important role in improving performance by optimizing resources properly, but efficient task scheduling still requires development. This research paper explores the application of reinforcement learning, specifically the Q-learning algorithms, for task scheduling. In this paper, we propose a multi-task scheduling framework which uses Q-learning algorithms. This model has two parts: in part 1, tasks are allocated to suitable servers in a dynamic fashion. In part 2, an improved Q-learning algorithm called ‘TSURLCC’ is used which assigns tasks, based on resources to each virtual machine available on the cloud server system. The machine gets a reward for assigning the tasks correctly, hence helping it to learn from past experiences. The proposed approach leverages the dynamic and unpredictable nature of cloud environments to adaptively schedule tasks, optimizing resource allocation, and minimizing execution time. Through extensive experimentation and analysis, we demonstrate improved task scheduling efficiency by using Q-learning algorithm. The findings highlight the potential of RL techniques to address the complexities of task scheduling in the context of cloud computing, paving the way for enhanced resource management and overall system performance.