AI Performance Modeling Architect for CPU/SoC Design

Actalent; Raleigh (NC); Hybrid; USD 200,000 - 500,000; Full time

Benefits offered by this job

Hybrid work in Raleigh

Job summary

Actalent, based in Raleigh, NC, is seeking a Performance Modeling Architect to develop and maintain modeling frameworks for evaluating architecture decisions in advanced computing platforms. The role spans hardware-software co-design, analyzing system performance, and guiding design tradeoffs for AI workloads.

You will build simulation tools and collaborate across hardware and ML teams to quantify impacts on performance, latency, and energy efficiency. Hybrid work in Raleigh is offered.

Qualifications

• Experience in computer or system architecture and performance modeling.
• Proficient in Python and C++ for modeling and simulation.
• Familiar with AI accelerators, CPUs/GPUs, and compute architectures.
• Experience modeling AI workloads or ML systems.
• Ability to identify performance bottlenecks and quantify tradeoffs.

Responsibilities

• Develop system-level performance models for AI workloads and compute architectures.
• Build simulation frameworks for interactions between AI models, runtimes, and hardware.
• Model SoC components: compute units, memory hierarchies, interconnects.
• Incorporate software/runtime behavior into models for realism.
• Analyze bottlenecks and evaluate architectural tradeoffs across the platform.
• Collaborate with hardware and software/ML teams to reflect real workloads.
• Evaluate performance, latency, throughput, and energy efficiency across apps.

Skills

Computer architecture

Performance modeling

Python

C++

AI accelerators

Bottleneck analysis

Tools

Simulation frameworks

Job description

Actalent, based in Raleigh, NC, is seeking a Performance Modeling Architect to develop and maintain modeling frameworks for evaluating architecture decisions in advanced computing platforms. The role spans hardware-software co-design, analyzing system performance, and guiding design tradeoffs for AI workloads.

You will build simulation tools and collaborate across hardware and ML teams to quantify impacts on performance, latency, and energy efficiency. Hybrid work in Raleigh is offered.