AI Performance Engineer (Kernel Systems)

TL;DR

AI Performance Engineer (Kernel Systems): Building the orchestration layer for AI compute with an accent on kernel optimization, memory hierarchy, and hardware execution efficiency. Focus on designing systems that enable high-performance AI workloads across heterogeneous accelerators at production scale.

Location: Must be based in San Francisco, CA (Onsite)

Salary: $250,000–$300,000 Base + Equity

Company

A rapidly growing AI infrastructure startup building the orchestration layer for AI compute, backed by $80m in Series A funding.

What you will do

• Optimize kernels for large-scale AI inference workloads.
• Improve memory movement, cache utilization, and execution efficiency.
• Tune performance for GPUs and emerging accelerator architectures.
• Develop kernel orchestration and execution planning systems.
• Analyze performance bottlenecks to maximize throughput and latency.
• Support execution across diverse hardware architectures.

Requirements

• Must be based in San Francisco, CA
• Experience building or optimizing performance-critical systems close to hardware.
• Strong understanding of GPU architecture and execution behavior.
• Deep knowledge of memory hierarchies, latency, throughput, and hardware efficiency.
• Strong software engineering fundamentals.
• Experience working on systems where performance and correctness are equally important.

Nice to have

• Experience with ROCm, Metal, or alternative accelerator backends.
• Experience optimizing AI inference or training workloads.
• Familiarity with occupancy tuning, latency hiding, and instruction-level parallelism.
• Experience with distributed or multi-GPU execution.
• Experience working alongside compiler, runtime, or systems teams.

Culture & Benefits

• Work at the intersection of kernel engineering, AI infrastructure, and hardware performance.
• Join a well-funded, high-growth startup with Fortune 500 deployments.
• Opportunity to define the execution layer for next-generation AI.
• Collaborative environment working with systems and runtime teams.