Senior Deep Learning Software Engineer, LLM Performance

Job Description

Senior role in NVIDIA’s deep learning inference team focused on analyzing and improving performance of large language and vision-language models across NVIDIA GPUs. The engineer works on GPU‑accelerated deep learning software and open‑source frameworks (e.g., TensorRT-LLM, vLLM, SGLang, Triton) to optimize large‑scale LLM and GenAI inference for datacenter deployments.

Responsibilities

• Analyze, profile, and optimize LLM, VLM, and GenAI workloads for low‑latency, high‑throughput inference across NVIDIA accelerators.
• Implement and tune inference, serving, and deployment algorithms within frameworks such as TensorRT-LLM, vLLM, SGLang, and Triton.
• Collaborate with research, hardware, and systems teams to identify cross‑stack optimizations that improve end‑to‑end model performance.
• Stay current with generative‑AI research and prototype emerging test‑time compute techniques (e.g., speculation, retrieval‑based refinement) for production use.

Benefits