Engineering Manager, Kubernetes Bare Metal Infrastructure

вчера

СШАLeadOnsite

kubernetesgpuinfinibandrocerdmacluster managementos imagingautomated provisioning

Manage and lead the engineering team responsible for Kubernetes bare-metal GPU infrastructure deployment and reliability at Together AI.

О компании

Together AI, the AI Native Cloud, is purpose-built for AI engineers. AI application developers get high-performance inference that scales reliably, fine-tuning and reinforcement learning for creating frontier-level specialized models, and pre-training at massive scale for fully custom intelligence, all around a marketplace of leading open models that teams can run, adapt, and own. Trusted by Cursor, Decagon, ElevenLabs, Salesforce, and Zoom, Together serves 400+ trillion tokens a month.

Обязанности

We are hiring an Engineering Manager/Tech Lead Manager to lead the team that runs Kubernetes on bare-metal GPU infrastructure at Together AI. You will own how thousands of GPUs across our data centers are provisioned, orchestrated and kept healthy for the customers training, fine-tuning and serving models on our platform.
Together AI is the AI Native Cloud: a full-stack platform for running open and custom generative AI models, from GPU clusters to serverless inference. Our research and engineering teams contribute widely used open-source work such as FlashAttention, and our infrastructure runs at the frontier of GPU hardware.
This team sits at the base of that stack. Every training run, inference endpoint and customer cluster depends on the reliability, utilization and speed of the bare-metal Kubernetes layer you will lead.
Own the bare-metal Kubernetes platform
Set the technical direction for how we run Kubernetes on bare metal across multiple data centers and regions.
Drive automated, zero-touch provisioning of GPU nodes, from firmware and OS imaging through cluster join and validation.
Own cluster lifecycle at scale: upgrades, capacity expansion, node health checks, automated remediation and decommissioning.
Make GPU scheduling, topology-aware placement and high-performance networking (InfiniBand, RoCE, RDMA) first-class in the platform.
Drive reliability and efficiency
Define and meet SLOs for cluster availability, node bring-up time and GPU utilization.
Invest in observability across hardware, networking, Kubernetes and workloads so failures are found before customers see them.
Cut the time from racked hardware to customer-ready capacity.
Lead and grow the team
Hire, manage and develop a team of infrastructure and platform engineers, setting a high bar for technical depth and operational ownership.
Run the team's planning, prioritization and delivery, balancing roadmap work against fleet growth and production support.
Build a healthy on-call culture: sustainable rotations, blameless postmortems and follow-through on remediation.
Work with data center operations, networking, hardware vendors, inference, training and product teams to plan capacity and launch new hardware generations.
Represent the platform's needs and trade-offs to leadership, and turn customer requirements into a clear roadmap.

Требования

8+ years in infrastructure, SRE or platform engineering, including 2+ years managing engineers.
Deep hands-on experience running Kubernetes in production at scale, ideally on bare metal rather than only managed services.
Solid grounding in Linux systems, networking and hardware: kernels, drivers, BIOS/BMC, PXE boot and out-of-band management.
Experience building automation for fleet provisioning and lifecycle with tools such as Cluster API, Metal3, Ironic, MAAS, Ansible or Terraform.
A track record of owning production systems with on-call, SLOs and incident response.
Strong technical judgment: you can review designs, debug alongside the team and make clear trade-offs.
Clear communication with engineers, leadership and external partners.
Nice to have
Experience operating large NVIDIA GPU clusters (H100, H200, B200, GB200) and the NVIDIA GPU and Network Operators, DCGM and NCCL.
Familiarity with InfiniBand, RoCE, SR-IOV and high-performance storage for AI workloads.
Experience with Slurm, Kueue, Volcano or other batch schedulers for training workloads.
Contributions to Kubernetes or related CNCF projects.
Experience at an AI lab, GPU cloud or hyperscaler.

Условия

We offer competitive compensation, startup equity, health insurance and other competitive benefits. The US base salary range for this full-time position is: $300,000 -350,000 + equity + benefits. Our salary ranges are determined by location, level and role. Individual compensation will be determined by experience, skills, and job-related knowledge.

Другое

Please read each role carefully and apply to the position best suited to you. Candidates are limited to 6 applications within a 90-day period, and applications beyond that limit will be automatically rejected. You may apply to additional roles 90 days after your last application.