Drimva

Arm

Principal Software Engineer, AI Inference Cloud

$263K–$355K

Seattle, Washington · Principal · Full-time · Posted Sep 9, 2026

Experience
12+ years, or equivalent proven impact.
Visa sponsorship
Relocation package, including visa sponsorship support, is available for candidates who require it.

About the job

Shapes the technical direction and builds reliable, scalable cloud services and Kubernetes capabilities for deploying and operating AI inference workloads.

What you'll do

  • Architect and develop AI inference services, Kubernetes controllers, and workload lifecycle capabilities.
  • Establish production practices and improve reliability, performance, scalability, and resource efficiency; lead production readiness and complex incident resolution.
  • Lead technical reviews and cross-team alignment, and mentor engineers while contributing hands-on.

What they ask for

  • 12+ years of experience, or equivalent proven impact, building distributed systems, cloud platforms, or production infrastructure.
  • Deep production experience with Kubernetes, including controllers, operators, scheduling, resource management, networking, and workload lifecycle management.
  • Strong software and production engineering skills, including hands-on programming in Go, C++, Rust, Python, or a similar language.
  • Track record leading complex technical initiatives hands-on, including architecture, implementation, debugging, mentoring, and influencing technical direction across teams.

Nice to have

  • Experience with AI infrastructure, model serving, or accelerator-backed workloads.
  • Familiarity with PyTorch, Ray, vLLM, SGLang, or TensorRT-LLM, or experience qualifying accelerators and tuning distributed workloads.

Pay

$262,700-$355,400 per year (base salary)

Tools

Kubernetes, Go, C++, Rust, Python, PyTorch, Ray, vLLM

AI focus

Develops and operates production AI inference infrastructure, including services and workload management; preferred experience includes model serving, accelerator-backed workloads, inference frameworks, and performance tuning.

Model-written summary · the original posting is authoritative

First seen Oct 28, 2026 · Last verified Nov 5, 2026 (UTC)