Worked on the infiniflow/ragflow repository to address a dataset isolation issue in Azure Blob storage, focusing on backend development and cloud storage integration using Python and Azure technologies. The solution involved updating connector modules to prepend a bucket prefix for all storage operations under both SPN and SAS authentication, ensuring that files from different datasets are stored under distinct prefixes and preventing cross-dataset file collisions. This approach aligned Azure Blob storage behavior with existing MinIO and S3 implementations, promoting consistency across storage backends. Manual verification confirmed that the fix effectively isolated datasets and resolved the previously reported collision bug.