Senior Site Reliability Engineer - GPU Infrastructure

Job not on LinkedIn

June 17

Apply Now

Receive Emails with Similar Jobs

Genmo

WebsiteLinkedInAll Job Openings

Artificial Intelligence • Media

Genmo is a company focusing on the development of cutting-edge AI video generation models. Their product, Mochi 1, is an open source video generation model that sets new standards in motion quality and realistic simulation according to the laws of physics. Genmo aims to solve fundamental problems in AI video technology, providing superior control over characters and settings through textual prompts. Mochi 1 is designed to create fluid, human-like actions and expressions, thereby advancing the capabilities of AI-generated video content. The company invites talent to join their team in building state-of-the-art open video models, highlighting roles for researchers, scientists, and engineers.

2 - 10 employees

🤖 Artificial Intelligence

📱 Media

📋 Description

• Own the design and day‑to‑day operation of GPU clusters that train and serve frontier generative models. • Lead production Kubernetes operations: GPU scheduling, cluster upgrades, multi‑cluster federation. • Define and implement Infrastructure‑as‑Code (Terraform, Helm, Ansible) and GitOps workflows with Argo CD or Flux. • Build CI/CD pipelines, automated testing, and rollout strategies for infra changes. • Develop an observability stack (Prometheus, Grafana, OpenTelemetry, eBPF) plus GPU telemetry with NVIDIA DCGM. • Optimize high‑performance networking (InfiniBand/RDMA) and debug perf bottlenecks. • Run and continuously improve the 24×7 on‑call rotation; lead post‑incident reviews. • Partner with researchers and engineers, communicate crisply, and ship with a high‑ownership mindset.

🎯 Requirements

• 3+ yrs SRE/DevOps in production; 2+ yrs managing large Kubernetes fleets. • Expert-level Kubernetes experience. • Proficient in Python and Bash and IaC tools (Terraform, Helm, Ansible). • Track record of shipping and operating large-scale infrastructure with high reliability and clear communication. • Multi-cluster / multi-cloud (AWS, GCP, Azure, bare-metal) production experience. • Hands-on with containerized GPU stacks (nvidia-container-toolkit, GPU Operator) • GPU schedulers such as Slurm or Kueue. • Familiarity with CI/CD tooling (GitHub Actions, BuildKit). • Prior work with distributed training, model-serving patterns, or other ML/GPU workloads.