Worked on the DS4SD/docling repository to stabilize the RAG LangChain Chunker Tokenizer, focusing on improving reliability and compatibility for end users. Addressed a key bug by updating the tokenizer implementation and adjusting dependency constraints, specifically loosening requirements to accommodate CVE fixes and support ASR tooling. Utilized Python for both data processing and machine learning tasks, ensuring that the environment remained secure and maintainable. Maintained thorough documentation and enforced DCO sign-off compliance to support collaborative development. The work emphasized release hygiene, improved commit traceability, and reduced future maintenance risk, contributing to a smoother upgrade path for the project.