TensorRT

TensorRT is NVIDIA's SDK for optimizing and running deep learning inference at maximum performance on NVIDIA GPUs. It supports quantization, custom plugins, and ONNX models, making it a foundational tool for deploying AI models in production.