Principal AI Silicon Architect

Principal AI Silicon Architect — Microsoft

Job #200044670 | Silicon Engineering | IC5 | Full-Time

What the role is: Design and lead the architecture of AI accelerator silicon for Microsoft's cloud infrastructure, working across hardware and software teams to optimize chips for modern AI workloads.

Key responsibilities:

• Define high-performance AI silicon architecture and specifications.
• Design and model AI compute/silicon IP blocks.
• Optimize architecture for current and future AI training and inference workloads.
• Work with hardware, AI software, firmware, kernel, design, verification, and emulation teams throughout the silicon development cycle.
• Guide projects from architecture through tape-out, bring-up, and validation.

Requirements:

• PhD + 3 years, Master's + 6 years, or Bachelor's + 8 years of technical engineering experience.
• Strong silicon architecture experience.
• Deep understanding of ML workloads, matrix multiplication, numerical formats, and parallel computing.
• Experience with Tensor/Vector processing architectures.
• Strong C/C++ and Verilog/SystemVerilog skills.
• Experience with bit-accurate C models and RTL design.

Preferred:

• 3+ years of silicon architecture experience.
• End-to-end silicon projects reaching successful tape-out and validation.
• Experience leading AI silicon IP from concept through bring-up.
• Strong knowledge of modern AI models for training and inference.

Work setup:

• 3 days/week in-office
• Less than 25% travel
• IC5
• U.S. salary: $142,800–$274,800/year
• SF Bay Area/NYC: $188,000–$304,200/year
• Export-control eligibility/citizenship verification required.

Bottom line: A highly senior AI chip architecture role focused on designing Microsoft's custom AI silicon and taking architectures all the way from concept to actual silicon.