During May 2025, Appy enhanced the quantization utilities in the pytorch/executorch repository by refactoring core components to integrate updated affine quantization methods. Leveraging Python and PyTorch, Appy aligned the quantization pipeline with torchao’s latest APIs, including choose_qparams_affine and quantize/dequantize_affine, to improve both efficiency and numerical accuracy. This work focused on backend development and machine learning, ensuring that quantization workflows remain consistent and adaptable to future changes. The changes were thoroughly documented and committed for traceability, reflecting a methodical approach to engineering. Appy’s contribution addressed the need for scalable, robust quantization processes without introducing new bug fixes during this period.