Senior C++ Training/Inference Engineer - High-Performance AI

SpaceXSI is hiring a Software Engineer, Training/Inference (C++) to design and optimize large-scale model serving systems used by Grok across millions of users. You will own distributed infrastructure from global KV cache to auto-scaling and batching, focusing on high performance and reliability. The role demands strong low-level programming in C/C++ or Rust, with experience in GPU kernels, quantization, and performance tuning to deliver fast, scalable inference engines.