Applying Scrum and Agile Practices to Large‑Scale Projects with Uncertain Requirements

Preface

The software market now emphasizes agile, big data, and cloud, but many customers cannot articulate concrete requirements and often demand a "benchmark" solution within a very short time and fixed budget, creating a classic "uncertain scope, limited resources, tight schedule" dilemma.

Project Development Process Design

Instead of starting from user requirements, the process begins with business research, followed by a Scrum‑based iterative development cycle. The team structure includes a planning manager, multiple feature teams, and a system team responsible for the overall architecture, with continuous integration, static code analysis, and code reviews. Unit testing is applied selectively to critical modules, achieving roughly 30% coverage.

Requirement Challenges

Requirements are handled asynchronously; detailed documents are replaced by a prioritized product backlog and collaborative clarification sessions involving PO, developers, and testers. The "three‑handshake" approach ensures that each story’s acceptance criteria are understood, recorded, and confirmed before development begins.

Defect Handling

Defects must be fixed immediately when discovered rather than batched at the end of a sprint. A dedicated bug‑fix specialist can be rotated among teams to keep story development flowing while maintaining high delivery value.

Story Splitting

Horizontal (layered) and vertical (feature‑oriented) splitting are discussed, with several practical techniques: by workflow, from simple to complex, CRUD operations, and common functional modules. Vertical splitting enables incremental delivery and isolated fixes.

Cross‑Team Coordination

Large projects often involve separate platform teams (big‑data, database, search) and business feature teams. Coordination mechanisms such as temporary reassignment of personnel or virtual teams are used to align story priorities and ensure end‑to‑end delivery.

Conclusion

While traditional Scrum works well for small teams, scaling to large, technically complex projects requires extended practices, explicit requirement clarification, immediate defect resolution, and flexible team structures to maintain continuous value delivery.