Contributed to the OpenDCAI/DataFlow repository by developing the RAREPipeline feature, which enables reasoning-intensive question generation using BM25 hard negative mining and LLM-based scenario creation. Enhanced the pipeline’s reliability by refactoring the RARE operator module to follow snake_case conventions and updating imports for improved readability. Addressed error handling in the Doc2Query operator by validating JSON object presence before decoding, preventing crashes from malformed API responses. Additionally, improved documentation quality in the langchain-ai/langchain repository by correcting spelling in Jupyter Notebook comments. Work primarily utilized Python and Jupyter Notebook, emphasizing data engineering, error handling, and code style consistency across projects.