This paper proposes a mixed integer non-linear programming (MINLP) for project scheduling and staff allocation problem with time-dependent learning effect considering task similarities for learning. Resources of a project, especially staffs, can learn from previous tasks if those tasks and the current task are similar. Considering similarities among tasks for each staff can produce a more realistic schedule for a project. The objective function of the proposed problem is to minimize project completion time. The learning effect in this paper is dependent on the time when the project staff starts to do that task. If a project staff does repetitive and/or similar tasks then that staff can get experience and do tasks faster than planned durations. Therefore, task sequence is essential to assign tasks to project staffs in order to minimize project completion time by considering task similarities in view of learning.