To address the challenges of sprint planning and sequential optimization in agile development projects, this paper presents a model for agile sprint clustering and optimization. Firstly, we build an activity similarity network and measure the tie strength between activities using social network analysis (SNA) and design structure matrix (DSM). Then, we build an “overlap-iteration” model to measure the interactions strength between activities. Based on thecoupling strength (i.e. the sum of tie strength and interaction strength) between activities, this paper uses a two-stage clustering method to cluster the process DSM to obtain agile sprints. Furthermore, we measure the rework probability and rework intensity between sprints from the rework probability and rework intensity between activities to build an agile sprint optimization method. Finally, an agile development project example is provided to illustrate the validity of the proposed models. The results demonstrate that our approach significantly reduces project duration, cost, and rework risk in agile development projects.