Worked on the adap/flower repository to enhance dataset partitioner reproducibility by clarifying and documenting the seed parameter’s role in initializing random number generation. Focused on Python-based refactoring to unify seed handling across partitioner classes, ensuring that dataset shuffling and other random processes behave deterministically for repeatable experiments. Updated documentation and code comments to reduce ambiguity around seed semantics, supporting more reliable benchmarking and collaboration across teams. Leveraged skills in Python, documentation, and code refactoring to improve maintainability and transparency in the codebase, enabling users to better understand and control the impact of randomness in dataset partitioning workflows.