Principal Compiler Developer (AI)

TL;DR

Principal Compiler Developer (AI): Developing new ML compiler capabilities in a modern stack from PyTorch through Triton down to CUDA or machine IR on leading GPUs and accelerators with an accent on performance-per-watt optimization for inference workloads. Focus on building and optimizing compiler infrastructure, lowering high-level workloads to efficient machine IR, and collaborating on HW-SW co-design.

Location: Santa Clara, CA; Remote; Washington, DC; San Francisco, CA; Denver, CO; Boston, MA; New York City, NY (US locations)

Salary: $145,000-$350,000 (various ranges)

Company

Early-stage startup backed by Tier-1 VC, founded by industry veterans, innovating AI infrastructure at chip and system level for order-of-magnitude better performance-per-watt in large-scale model inference.

What you will do

• Build and optimize ML compiler infrastructure across the stack from PyTorch/Triton to low-level code generation.
• Lower PyTorch/Triton workloads to efficient machine IR for GPUs and new AI accelerators.
• Optimize performance for CUDA/ROCm on leading-edge GPUs and XPU hardware.
• Collaborate on hardware-software co-design in a fast-moving startup environment.
• Lead or contribute to a small team developing compiler capabilities (depending on experience).

Requirements

• PhD + 5 years (or equivalent) in compilers, ML systems, or related fields
• Strong compiler fundamentals and performance optimization skills
• Experience with PyTorch, Triton, and low-level code generation

Nice to have

• GPU expertise (CUDA/ROCm)
• HW-SW co-design experience