Worked on the livekit/agents repository to deliver text-to-speech model support for the lightning-v3.1 release, focusing on robust API and back end development using Python. Developed and integrated updated routing logic that enables the system to correctly handle requests for both lightning-v2 and lightning-v3.1 models, ensuring accurate model selection for audio synthesis. Addressed endpoint routing issues by implementing and documenting a targeted fix, which improved the reliability of model dispatch within the TTS system. The work emphasized clear documentation and maintainable code, contributing to a more flexible and extensible text-to-speech synthesis pipeline for future model integrations.