Triton inference server publicly released on: 2018-11 by NVIDIA — formerly TensorRT Inference Server.