Senior GPU Kernel Engineer for High-Performance AI Inference

CoreWeave; Sunnyvale (CA); On-site; USD 182,000 - 242,000; Full time

Benefits offered by this job

Medical, dental, and vision insurance

401(k) with employer match

ESPP

Flexible PTO

Catered lunch

Job summary

CoreWeave is hiring a Senior Engineer for its Benchmarking & Performance team to write, profile, and optimize GPU kernels on the LLM inference path. You will improve latency and throughput and collaborate with product, orchestration, and hardware teams to achieve strict P99 SLAs at scale.

You will lead kernel-level performance work, publish rigorous benchmarks, and mentor engineers while shaping the benchmarking workflow and end-to-end performance stories for model serving stacks.

Qualifications

• 5+ years of experience building high-performance computing, GPU/accelerator software, or performance-critical systems.

Responsibilities

• Author, profile, and optimize CUDA kernels on the critical path of LLM inference.
• Optimize for hardware: tensor cores, occupancy, memory hierarchy, and overlap of compute with data movement.
• Benchmark rigorously with microbenchmarks and roofline analyses and translate kernel wins to end-to-end gains.
• Lead design reviews and drive architecture within the team; mentor junior engineers.

Skills

CUDA kernels

C++

Python

GPU architecture

Nsight Compute/Systems

Tools

Nsight Compute

Nsight Systems

vLLM/TensorRT-LLM

llm-d

SGLang

Job description

CoreWeave is hiring a Senior Engineer for its Benchmarking & Performance team to write, profile, and optimize GPU kernels on the LLM inference path. You will improve latency and throughput and collaborate with product, orchestration, and hardware teams to achieve strict P99 SLAs at scale.

You will lead kernel-level performance work, publish rigorous benchmarks, and mentor engineers while shaping the benchmarking workflow and end-to-end performance stories for model serving stacks.