During April 2025, Jun Li contributed to the eosphoros-ai/DB-GPT repository by enhancing both data integrity and NLP extensibility. Jun addressed a bug in the chat history system, updating the backend Python logic to ensure app_code was reliably stored and associated with each conversation, which improved traceability for auditing. Additionally, Jun integrated the SiliconFlow embedding proxy, introducing a new embedding generation class and updating TOML-based system configuration to support external embedding models. This work leveraged skills in API integration and backend development, enabling the system to utilize external embeddings and reducing vendor lock-in for future natural language processing experiments.