Hardware Benchmarking Engineer – AI Accelerators

Artificial Analysis, Inc.; San Francisco (CA); On-site; USD 150,000 - 210,000; Full time

Benefits offered by this job

Equity

Job summary

Artificial Analysis, Inc. is seeking a Member of Technical Staff (Hardware) to design and run AI hardware benchmarks for GPUs, TPUs and custom silicon in San Francisco.

You will develop and extend benchmarking stacks and work closely with chipmakers to shape methodology and standards. The role requires 3+ years in AI accelerators, strong Python/data-analysis skills, and deep knowledge of inference economics.

Qualifications

• 3+ years of professional experience in AI hardware or accelerators.
• Experience at an AI chip company (e.g. NVIDIA, Cerebras, SambaNova, Groq or similar).
• Proficiency in Python and data analysis.
• Deep familiarity with AI accelerators and the inference software stack (CUDA, TensorRT, vLLM or similar).
• Understanding of inference economics: cost-per-token, utilization, deployment trade-offs.
• Genuine interest and knowledge of Frontier AI.

Responsibilities

• Benchmark AI accelerators across GPUs, TPUs and custom silicon, measuring throughput, latency and price-performance.
• Develop frameworks to measure AI hardware performance and efficiency, extending AA-AgentPerf and benchmarking tools.
• Analyze inference economics and translate findings for deployment and procurement decisions.
• Collaborate with leading chipmakers to benchmark latest silicon and refine methodology.
• Produce reports and data visualizations communicating hardware performance to technical and non-technical audiences.
• Adopt an AI-native workflow using cutting-edge AI tools to maintain a competitive edge.

Skills

Python

Data analysis

Analytical thinking

Inference economics

Frontier AI knowledge

AI accelerator experience

Tools

CUDA

TensorRT

vLLM

Job description

Artificial Analysis, Inc. is seeking a Member of Technical Staff (Hardware) to design and run AI hardware benchmarks for GPUs, TPUs and custom silicon in San Francisco.

You will develop and extend benchmarking stacks and work closely with chipmakers to shape methodology and standards. The role requires 3+ years in AI accelerators, strong Python/data-analysis skills, and deep knowledge of inference economics.