Yash Nankani developed a mixed-precision quantization feature for the hpcaitech/TensorRT-Model-Optimizer repository, focusing on enabling configurable accuracy and performance trade-offs for machine learning model deployment. Using Python and leveraging data processing and quantization techniques, Yash implemented support for INT4 and INT8 quantization, allowing users to specify 8-bit layers via new CLI options. The technical approach included enhancements to precision mapping and scaling functions, ensuring performance gains while maintaining model accuracy. This work expanded deployment flexibility for resource-constrained environments, reducing model size and improving inference throughput. The feature was delivered with validated tests and no major bugs reported during the period.