Member of Technical Staff - AI Hardware & System Performance
Core
Quantitative analysis of AI hardware and systems to build first-principles performance models, evaluate architecture trade-offs, and translate technical differences into measurable performance and economic outcomes.
Role type
Senior IC AI hardware analyst (systems & economics)
Builds
Industry models on accelerator shipments, datacenter demand, GPU TCO, and AI infrastructure economics
Domain
Semiconductor industry + AI hardware systems
Deliverable
production ML models | dashboards & analysis
Required skills
first-principles hardware modeling, arithmetic intensity analysis, roofline performance evaluation, memory bandwidth analysis, semiconductor manufacturing economics, power/cooling constraints analysis, TCO calculation, technical due diligence, quantitative reasoning, datacenter infrastructure knowledge
Preferred skills
AI inference frameworks (vLLM, SGLang, TensorRT-LLM), GPU programming (CUDA, Triton, HIP), transformer inference mathematics, large cluster operations, datacenter networking, heterogeneous hardware benchmarking
Technologies
vLLM, SGLang, TensorRT-LLM, PyTorch, JAX, CUDA, Triton, HIP, NVIDIA GPUs, AMD GPUs, TPUs, HBM, 2.5D integration, 3D stacking
Responsibilities
Build first-principles performance models for LLM inference and training; Analyze arithmetic intensity, roofline performance, and latency-throughput trade-offs; Evaluate AI hardware across the full technology stack; Assess architecture and system design trade-offs; Connect semiconductor manufacturing decisions to hardware performance and cost; Analyze process-node choices and silicon economics; Examine new chip and system announcements; Develop independent views on real-world performance; Validate internal models against benchmarks; Translate technical analysis into economic metrics; Publish detailed research on AI accelerators and systems; Act as a technical authority during client calls
Seniority
Senior, hands-on IC
