Large-scale software development organizations operate in environments characterized by multiple parallel value streams, distributed teams, and dependencies on external stakeholders. Although agile software development enables flexibility, existing project management approaches often provide insufficient support for long-term resource planning across complex portfolios. This paper proposes and validates a data-driven resource planning methodology that bridges project management and agile product development while preserving agile principles and improving planning predictability. The methodology integrates four complementary elements: net capacity estimation, historical data analysis, planning policies, and a planning matrix for allocating resources across multiple value streams. Historical data extracted from work management systems such as Jira are continuously analyzed to estimate task effort and delivery timelines, enabling iterative calibration of planning parameters. The methodology was developed and evaluated through its implementation in a large-scale industrial software development organization. The evaluation demonstrated improved planning accuracy, more effective utilization of development capacity, better coordination of parallel projects, and greater transparency for stakeholders. Continuous refinement of planning parameters based on empirical project data reduced discrepancies between planned and actual workloads while supporting more predictable delivery planning. The results indicate that the proposed methodology provides a practical and scalable framework for resource planning in complex software development environments and establishes a foundation for future research on integrating predictive analytics and intelligent decision-support mechanisms to further enhance planning accuracy and resource allocation.

Data-driven resource planning agile project management large-scale software development