Senior Staff AI Accelerator Performance Architect
$175k - $275k • Sunnyvale, CA • FullTime
Posted 1mo ago
Job Location
Sunnyvale, CA
Tech Stack
Remote Work Policy
On-site
Employment Type
FullTime
Categories
Applied AI Engineer
About the job
Cerebras Systems is seeking a Senior Staff AI Accelerator Performance Architect to guide the evolution of their next-generation AI systems. This role involves connecting real-world AI workloads to architectural behavior, identifying critical bottlenecks, quantifying potential improvements, and influencing hardware and software roadmaps through rigorous performance analysis. The ideal candidate will possess deep knowledge of hardware architecture, gained through experience in hardware, compiler, kernel, or system performance work, and thrive at the intersection of applications, kernels, architecture, and system performance. This position focuses on performance evaluation and architectural recommendations, rather than production RTL design.
Responsibilities
- Own and evolve performance models and methodologies for next-generation AI accelerator and system architectures.
- Build and extend analytical, simulation-based, or trace-driven models across various workloads and product generations.
- Analyze AI workloads to understand resource utilization (time, bandwidth, compute, capacity).
- Identify hardware and software bottlenecks and quantify opportunities for performance, latency, throughput, utilization, and energy efficiency improvements.
- Evaluate proposed architectural features and their expected performance impact.
- Study the mapping of models and kernels onto compute, memory, and communication architectures.
- Collaborate with architecture, compiler, kernel, runtime, and systems teams on alternative mappings and optimizations.
- Develop workload projections and competitive performance analyses.
- Translate complex performance results into clear architectural and product recommendations.
- Improve modeling methodology, validation, and correlation with hardware measurements.
- Help define representative workloads, performance targets, and success criteria for future products.
Requirements
- 7+ years of experience in performance analysis, modeling, or architecture exploration for HPC systems (CPUs, GPUs, AI accelerators).
- Strong understanding of hardware architecture from hardware, compiler, kernel, runtime, or system performance work.
- Experience developing performance models using Python, C++, or similar languages.
- Solid understanding of processor architecture, memory systems, interconnects, parallel execution, and hardware constraints.
- Ability to analyze performance from kernel-level behavior to end-to-end application/system performance.
- Experience profiling workloads, forming performance hypotheses, and validating them with data.
- Understanding of how software mapping and programmability affect hardware performance.
- Ability to clearly communicate modeling assumptions, uncertainty, bottlenecks, and recommendations.
- MS or PhD in Electrical Engineering, Computer Engineering, Computer Science, or equivalent practical experience.
Benefits
- Bonus
- Equity