Senior AI Inference Performance Engineer (CUDA/LLM/VLM)

NVIDIA AI; Santa Clara (CA); On-site; USD 180,000 - 300,000; Full time

Benefits offered by this job

Equity

Generous Benefits Package

Job summary

Lead the end-to-end performance analysis and optimization of LLM and VLM inference on NVIDIA GPU-accelerated systems. Develop performance-critical kernels and contribute improvements to open-source inference engines to reduce latency and increase throughput.

Requires over 6 years of experience in full-stack AI inference performance with strong programming skills in Python, C++, or Rust and expertise in CUDA.

Qualifications

• Requires 6+ years in full-stack AI inference performance.
• Strong programming skills in Python, C++, or Rust and expert CUDA.
• Degree in Computer Science or Computer Engineering required.
• Deep knowledge of GPU architecture and profiling tools.

Responsibilities

• Lead end-to-end performance analysis and optimization of LLM and VLM inference on NVIDIA GPU-accelerated systems.
• Develop performance-critical kernels and contribute improvements to open-source inference engines to reduce latency and increase throughput.

Skills

Full-stack AI

Python

C++

Rust

CUDA

GPU Architecture

Profiling Tools

Nsight Systems

Nsight Compute

PyTorch Profiler

TensorRT-LLM

vLLM

Triton

CUTLASS

Distributed Systems

Quantization

Education

Bachelor's degree in Computer Science or Computer Engineering

Tools

Nsight Systems

Nsight Compute

PyTorch Profiler

TensorRT-LLM

vLLM

Triton

CUTLASS

Job description

Lead the end-to-end performance analysis and optimization of LLM and VLM inference on NVIDIA GPU-accelerated systems. Develop performance-critical kernels and contribute improvements to open-source inference engines to reduce latency and increase throughput.

Requires over 6 years of experience in full-stack AI inference performance with strong programming skills in Python, C++, or Rust and expertise in CUDA.