Senior SoC Architect for AI Hardware & System Design

Voltai; United States; On-site; USD 180,000 - 260,000; Full time

Job summary

Voltai is seeking an experienced SoC Architect to define architecture for SoCs, subsystems, and IP blocks, from requirements through specs used by RTL teams. You will model performance, power, and area and guide trade-offs before RTL is written on real silicon programs with top customers.

You will collaborate with ML researchers to teach models how architects reason about design trade-offs and work directly with customers to translate product requirements into silicon.

Qualifications

• 8+ years of experience in SoC, CPU, GPU, accelerator, or networking architecture.
• Track record of taking architectures from spec to tapeout.
• Strong performance modeling skills using SystemC, C++, Python, or equivalent.
• Deep understanding of memory systems, interconnects, cache coherence, and PPA trade-offs.
• Clear technical writing and the ability to lead reviews across teams.

Responsibilities

• Own architecture and microarchitecture for SoCs, subsystems, and IP blocks, from requirements through specification.
• Build performance, power, and area models to drive trade‑off decisions before RTL is written.
• Define memory hierarchies, interconnects, and interfaces, and lead architecture reviews.
• Write specifications that RTL, verification, and physical design teams can execute against.
• Partner with ML researchers to teach models how expert architects reason about design trade‑offs.
• Work directly with customers to translate their product requirements into silicon.

Skills

SoC architecture

CPU architecture

GPU architecture

accelerator architecture

memory systems

performance modeling

PPA trade-offs

technical writing

interconnects

Tools

SystemC

C++

Python

Job description

Voltai is seeking an experienced SoC Architect to define architecture for SoCs, subsystems, and IP blocks, from requirements through specs used by RTL teams. You will model performance, power, and area and guide trade-offs before RTL is written on real silicon programs with top customers.

You will collaborate with ML researchers to teach models how architects reason about design trade-offs and work directly with customers to translate product requirements into silicon.