During July 2025, this developer integrated DeepSeek and BAAI/bge-m3 embedding models into the HKUDS/VideoRAG repository, enabling advanced video processing and content-based querying. They implemented end-to-end embedding-based retrieval, allowing VideoRAG to support scalable video search and analytics. The work involved API integration and asynchronous programming in Python, with careful attention to data handling and video processing workflows. By merging these features, the developer established a foundation for improved search accuracy and content discovery within the project. The depth of the integration demonstrates a solid understanding of embedding models and their application to real-world video retrieval challenges.