Senior CUDA Kernel Engineer for High-Performance Inference

CoreWeave; Bellevue (WA); On-site; USD 182,000 - 242,000; Full time

Benefits offered by this job

Medical, dental, and vision insurance

Company-paid Life Insurance

Disability insurance

401(k) with employer match

Flexible PTO

Catered lunch

Job summary

CoreWeave is seeking a Senior Engineer for its Benchmarking & Performance team to write, profile, and tune CUDA kernels on the inference path. You will enhance latency, throughput, and reliability across the stack and work across product, orchestration, and hardware teams.

You will mentor engineers, lead design reviews, and help deliver industry-leading MLPerf results. A strong CUDA background and GPU architecture expertise are required for this critical role at CoreWeave.

Qualifications

• 5+ years of experience building high-performance computing, GPU software, or performance-critical systems.
• Hands-on CUDA experience with custom kernel writing and optimization.
• Deep understanding of GPU architecture including tensor cores and memory hierarchy.
• Proficient in C++ and Python; able to read/write low-level, performance-sensitive code.
• Familiarity with model-serving stacks and kernels that dominate inference cost.

Responsibilities

• Author, profile, and optimize CUDA kernels on the path of LLM inference.
• Tune for tensor cores, occupancy, memory coalescing, and data movement overlap.
• Prototype and ship kernels using DSLs and compilers while maintaining performance.
• Benchmark end-to-end latency and throughput, validate with reproducible results.
• Lead design reviews, drive architecture, and mentor junior engineers.
• Ensure reproducible benchmarking and well-documented processes.

Skills

CUDA programming

C++

Python

GPU architecture

communication

Tools

Nsight Compute/Systems

CUDA kernel development

Job description

CoreWeave is seeking a Senior Engineer for its Benchmarking & Performance team to write, profile, and tune CUDA kernels on the inference path. You will enhance latency, throughput, and reliability across the stack and work across product, orchestration, and hardware teams.

You will mentor engineers, lead design reviews, and help deliver industry-leading MLPerf results. A strong CUDA background and GPU architecture expertise are required for this critical role at CoreWeave.