Worked on the NVIDIA-NeMo/Export-Deploy repository to enhance the usability and clarity of the deploy_inframework_triton.py script’s command-line interface. Focused on improving documentation and scripting by updating help text to accurately reflect parameter functions, such as clarifying that --num_nodes refers to the number of nodes rather than GPUs and redefining --context_parallel_size for better alignment with context parallelism. This targeted update aimed to reduce user confusion and streamline onboarding for new users deploying Triton-based export pipelines. The work was implemented in Python, emphasizing maintainability and user experience without introducing new features or bug fixes during the development period.