Worked on the resemble-ai/chatterbox repository to deliver a major upgrade to the Voice Conversion feature, focusing on improved model loading and developer onboarding. Introduced safetensor format support within the Python-based deep learning workflow, which enhanced loading efficiency and flexibility for audio processing tasks. Developed new example scripts and expanded WAV sample sets to streamline testing and demonstration processes, supporting faster deployment and easier onboarding for new contributors. Added a watermark extraction workflow to the demo suite, enabling better compliance and metadata handling. All work was closely aligned with project tracking to facilitate smoother deployment and cross-team collaboration within the repository.