Tether

Technical Lead - GPU Infrastructure

Rabat

CDI / Temps pleinTélétravailLicence / Bac+3AnglaisInformatique / IngénierieInformatique / IT

L’essentiel du poste

Résumé par Postule AI

Technical Lead role at Tether Data leading the GPU infrastructure platform (Cosmic AC). Requires 8+ years of hands-on engineering with 3+ years leading infrastructure platform teams. Must have deep expertise in Slurm at scale, bare-metal GPU operations, Kubernetes production systems, HPC storage, high-performance interconnects and Linux systems. This is a hands-on leadership position managing 12 distributed engineers with a fixed 6-month delivery window, combining architecture ownership, team management and direct technical work.

Vos missions

  • Own platform architecture end-to-end including architecture proposals, designs and baseline maintenance
  • Lead and line-manage distributed team of 12 engineers across backend, frontend, DevOps, QA and documentation
  • Design and operate managed Slurm service for research users with controller, accounting and node health management
  • Own Kubernetes control plane and GPU enablement including cluster bootstrap, NVIDIA GPU Operator and day-2 operations
  • Architect managed inference at scale with multi-GPU parallelism, autoscaling and endpoint reliability
  • Establish observability, incident response and sustainable on-call models for the platform

Profil recherché

  • You have 8+ years hands-on engineering with proven track record leading infrastructure platform teams
  • You have deep operational experience running Slurm and Kubernetes in production at scale
  • You combine technical depth in GPU systems, HPC and Linux with people leadership skills
  • You excel at written communication and can articulate architecture decisions and constraints clearly
  • You thrive in hands-on technical leadership roles that require both code review authority and team management

Généré automatiquement à partir de l’offre. Postule AI peut se tromper : vérifiez auprès de l’employeur.

About the Role

Lead the architecture and delivery of Tether Data's GPU compute and managed inference platform (Cosmic AC). This hands-on infrastructure leadership role involves owning the full stack on bare-metal GPU infrastructure, managing a distributed team of twelve engineers across backend, frontend, DevOps, QA and documentation, and serving as the primary technical interface to infrastructure partners.

Main Responsibilities

  • Own platform architecture end-to-end: architecture proposals, high-level and low-level designs, and baseline maintenance
  • Lead and line-manage a distributed team across backend (Node.js), frontend (React), DevOps, QA and documentation with engineering standards and performance reviews
  • Design, build and operate a managed Slurm service for research users including controller, accounting, node onboarding, driver and CUDA lifecycle management
  • Own Kubernetes control plane and GPU enablement: cluster bootstrap, NVIDIA GPU Operator, VM-based GPU isolation (KubeVirt and VFIO), and day-2 operations
  • Architect managed inference at scale with multi-GPU and multi-node parallelism, autoscaling, request routing and endpoint reliability
  • Establish observability and operations: metrics, logging, alerting, SLOs, incident response and sustainable on-call models

Required Qualifications

  • Eight or more years of hands-on engineering with at least three years leading teams that build and operate infrastructure platforms
  • Hands-on Slurm operation at scale: slurmctld, slurmdbd, partitions, QoS, accounting, node health scripting and upgrades
  • GPU fleet operation on bare metal: NVIDIA driver and CUDA lifecycle, Fabric Manager, DCGM-based health monitoring, MIG and node acceptance
  • High-performance interconnects expertise: InfiniBand fabric configuration, RDMA, SR-IOV and multi-node NCCL performance diagnostics
  • Deep Linux systems knowledge: kernel modules, drivers, PCIe passthrough, vfio-pci, cgroups, namespaces and compute workload tuning
  • Production Kubernetes operation: control plane, upgrades, CNI and CSI, operators, custom controllers and multi-tenancy design
  • HPC storage and data movement: shared filesystems (VAST, Lustre, NFS), node-local NVMe caching and large-scale model distribution
  • Observability and operations: Prometheus, Grafana, Loki, SLOs, incident response and post-incident review

Additional Requirements

  • Working fluency in JavaScript and Node.js sufficient to review control plane, CLI and worker services with authority
  • Shipped platform experience with real users: multi-tenant IaaS/PaaS or research computing service with resource isolation and usage metering
  • Leadership that stays in the code: people management across time zones, written architecture decisions and ability to communicate constraints clearly
  • Excellent written and spoken English for partner and leadership communication
  • Location: Fully remote, based between UTC and UTC+5:30 to overlap Europe and India team working hours; occasional travel to partner sites

Desirable Experience

  • Slurm operators on Kubernetes (Soperator, Slinky) or Kubernetes-native schedulers (Kueue, Volcano, KAI, Kubeflow Trainer)
  • Modern serving stacks: vLLM, SGLang, TensorRT-LLM with parallelism strategies and GPU memory planning
  • VM and container isolation for multi-tenant GPU compute and confidential computing technologies
  • Cluster API, kubeadm, Cilium, infrastructure as code and GitOps practices
  • Experience with GPU clouds, national HPC centres or AI lab platform teams

Mise en forme par Postule ·

Signaler un problème