Jax, Linux System Administration, Bash, Cuda, Pytorch, Docker, Kubernetes, Python, TensorRT-LLM, inference benchmarking, RDMA, infiniband, Hugging Face Transformers, GPUDirect, Go, high-performance networking technologies, SGLang, vLLM, AI model training, performance characterization, NCCL, AI ML frameworks

triton , Distributed Systems, Hpc, Cuda, Jax, Performance Engineering, Machine Learning, XLA, Model Parallelism, Tensor Parallelism, Pipeline Parallelism, FSDP, NCCL, GPU Optimization, Kernel Optimization, Mixed Precision, FP8, Checkpointing, MFU, ML Infrastructure, AI/ML, Llm