Developed an auto-detection feature for MLX-format quantization configurations in the yhyang201/sglang repository, streamlining the deployment of pre-quantized machine learning models on Apple Silicon. Leveraged Python to enhance backend logic, enabling the system to recognize and validate MLX-specific quantization settings without requiring explicit user input. Focused on improving quantization configuration handling, the work reduced setup friction and improved compatibility for Apple Silicon users. Applied skills in backend development, machine learning, and quantization, with thorough unit testing to ensure correctness. These enhancements laid the foundation for broader MLX format support and accelerated model deployment workflows for developers and users.