Member of Technical Staff, Local Inference & Kernels

📋 Description

• Own end-to-end performance: profile latency, throughput, peak memory, and energy use across
• Build kernels where they matter: optimize critical operations such as matrix multiplication
• Make low-precision inference useful: evaluate weight and KV-cache quantization, mixed precision
• Improve the runtime: reduce allocation and kernel-launch overhead, manage caches, and schedule
• Co-design for consumer hardware: help researchers assess architecture and context-length choices
• Build performance infrastructure: maintain reproducible benchmarks and numerical checks across

🎯 Requirements

• Deep experience in ML inference, GPU programming, or numerical computing, with concrete examples of
• Strong C++ skills and hands-on experience writing kernels in Metal, CUDA, Triton, or a comparable
• Working understanding of GPU architecture: memory hierarchies, bandwidth, SIMD execution, register
• Experience optimizing matrix multiplication, attention, or similarly demanding operations
• Understanding of quantization and mixed precision, and the ability to measure effects on memory
• Strong profiling instincts: trace an application-level slowdown to the responsible operation

🎁 Benefits

• Comprehensive health, dental, and vision coverage.
• Flexible PTO.
• Relocation support as needed.
• Meaningful equity.