AI Accelerator Performance Engineer

MatX Inc.; Mountain View (CA); Hybrid; USD 120,000 - 600,000; Full time

Benefits offered by this job

Time off

Health insurance

Financial Wellbeing

Professional Development

Team Meals

Commute on Us

MatX E[x]tras

Cell & Internet Reimbursement

Remote Home Setup

Mental Wellbeing

Parental Leave

Job summary

MatX Inc. is building cutting-edge AI accelerator modeling to push frontier models toward higher efficiency and performance. The role focuses on translating architectures into performance models, maintaining the modeling framework, and automating tensor layouts for frontier LLMs.

You will collaborate with ML, software, and architecture teams to support experiments, develop tools and compilers, and advance the company's AI stack.

Qualifications

• Strong background in performance modeling.
• Strong working knowledge of latest LLM model architectures.
• Experience with computer architecture, particularly AI accelerators.

Responsibilities

• Translate AI accelerator architecture into performance models.
• Maintain modeling framework capabilities and releases.
• Automate creation of detailed tensor layouts for frontier LLMs.
• Create tools, compilers, and automations to integrate with the broader organization tech stack.
• Collaborate with ML, Software, and Architecture teams to support experiments and workflows.

Skills

Performance modeling

LLM architectures

AI accelerators

Job description

MatX Inc. is building cutting-edge AI accelerator modeling to push frontier models toward higher efficiency and performance. The role focuses on translating architectures into performance models, maintaining the modeling framework, and automating tensor layouts for frontier LLMs.

You will collaborate with ML, software, and architecture teams to support experiments, develop tools and compilers, and advance the company's AI stack.