Senior Software Engineer - GPU Kernel Authoring & Optimization

📋 Description

• Write, profile, and optimize CUDA kernels on the LLM inference path.
• Tune GPU kernels for throughput, latency, and memory bandwidth.
• Benchmark with microbenchmarks and roofline analyses; prove end-to-end gains.
• Lead design reviews; decompose work into milestones; mentor engineers.
• Collaborate with product, orchestration, and hardware teams to improve the inference stack.
• Support end-to-end MLPerf Inference (and Training) benchmark runs.

🎯 Requirements

• 5+ years in HPC, GPU software, or performance-critical systems.
• Hands-on CUDA experience; written/optimized custom kernels.
• Deep GPU architecture knowledge: tensor cores, occupancy, memory.
• Proficient in C++ and Python; low-level performance code.
• Familiar with model-serving stacks: vLLM, TensorRT-LLM, llm-d, SGLang.
• Strong communication and cross-functional collaboration.

🎁 Benefits

• Medical, dental, and vision insurance - 100% paid for by CoreWeave
• 401(k) with generous employer match
• Flexible PTO
• Tuition Reimbursement
• Employee Stock Purchase Program (ESPP)
• Mental Wellness Benefits through Spring Health