Team: Infra Reliability · SF Bay Area / Remote (US)

You'll own the GPU infrastructure Luma's research and product run on — thousands of NVIDIA and AMD GPUs across on-prem and multi-cloud (AWS and OCI). As a Senior SRE, you keep training and inference clusters reliable and fast, and you help redesign them for the next level of scale.

This is a hands-on, close-to-the-metal role for a first-principles Linux engineer. You'll be the final escalation for the hardest GPU, networking, and kernel-level failures, sometimes debugging directly with NVIDIA. It fits someone who thrives on low-level problems in a fast, less-structured environment. If you want a narrow, well-bounded ops role, this isn't it.

What You'll Own

• Take end-to-end ownership of production GPU clusters for training and inference across AWS and OCI, keeping them highly available and performant.

• Join critical re-architecture sessions to redesign systems for higher efficiency and scale.

• Tune Linux performance deeply, at the OS and kernel level.

• Build automation in Python, Go, or Bash to manage, monitor, and self-heal infrastructure without heavy toil.

• Serve as the final escalation for the hardest GPU, networking (InfiniBand/RDMA), and system failures, working with vendors like NVIDIA.

• Help achieve and maintain security certifications (SOC 2 Type 1 & 2, ISO) with strong infrastructure security practices.

First 90 Days

One way the first 90 could unfold.

• Days 1–30 — Immerse & Diagnose: Learn the current clusters across on-prem, AWS, and OCI, and where reliability and performance hurt most.

• Days 30–60 — Ship & Validate: Take ownership of a production cluster and ship automation or tuning that measurably improves availability or performance.

• Days 60–90 — Scale & Systemize: Contribute to the next-gen re-architecture and harden security and compliance practices.

What You Bring

• 5+ years as an SRE, production, or infrastructure engineer in a fast-paced, large-scale environment.

• Deep, hands-on Linux expertise, containerized systems, and low-level performance debugging.

• Working experience with Terraform, Airflow, and Ray.

• Strong experience with AWS or OCI.

• Practical experience with high-performance networking (InfiniBand, RDMA, or RoCE).

• Working knowledge of security best practices and compliance frameworks like SOC 2 and ISO.

• Comfort in a less-structured, fast-paced environment.

Nice to Have

• Deep expertise with GPU tooling for NVIDIA and AMD (DCGM, ROCm).

• Experience managing large-scale GPU clusters for AI/ML training or inference.

• Familiarity with Kubernetes or orchestration frameworks like Ray.

• Deep expertise in data pipelines and infrastructure.

About Luma: Luma's mission is to build unified general intelligence that can generate, understand, and operate in the physical world. We believe multimodality is critical for intelligence — the next step beyond language models comes from vision. Luma is an equal opportunity employer.