Agent-based systems are deployed in distributed data and control environments where resources usage are claimed to be more efficient. Researches have shown that intelligent agents offer excellent performance in distributed environments influenced by their autonomous or semi-autonomous actions with minimal human intervention. Assigned with roles, these agents are thus viewed as human-like societies that are capable of planning, coordinating, and executing tasks to achieve their goals. In a similar fashion, we model agents in our work based on the peculiarities of human society to achieve optimum performance in a dynamic multi-skills environment. Due to time and task constraints, agents are assigned with multiple roles and perform various other tasks. In this paper, we discuss a framework for dynamic task assignment and delegation in handling multi-task problems for a society of agents to achieve a common goal in allocated time. To test the framework, we adopt and animate the Final Examination Paper Preparation (FEPP) process in a multi-agent system that involves a number of faculty branches where multiple and dynamic task delegations are possible.