Senior Software Engineer, Quantized Inference

Accelerate the discovery and deployment of efficient quantized and sparse inference “recipes” for LLMs. Translate recipe specifications into performant inference-engine implementations across vLLM, TRT-LLM, and SGLang—writing Triton kernels, inserting quantize/dequantize nodes, and handling MoE scaling correctly. Own model export pipelines (ModelOpt, Megatron-LM, HuggingFace), build benchmarking harnesses and numerics debugging tools, and improve developer productivity through CI and build system enhancements.