In August 2025, Michael Howto developed scalable embedding capabilities by integrating Gemini Embeddings into both the cloudwego/eino-ext and coze-dev/coze-studio repositories. He implemented new embedding components and initialization logic in Go, focusing on robust API integration and backend development. His work included configuration management, embedding logic, and basic test coverage, ensuring consistent deployment across cloud services. By standardizing embedding configuration and environment variables, Michael enabled smoother cross-team adoption and improved CI reliability. This foundation supports downstream machine learning features such as search and similarity, demonstrating depth in embedding models and laying groundwork for future ML-powered enhancements within both projects.