Technical Lead - GPU Infrastructure

About the role

Lead the design, implementation and delivery of Tether Data's GPU compute platform, overseeing a distributed team of engineers and acting as the primary technical liaison with infrastructure partners.

Key responsibilities

• Own end‑to‑end platform architecture, from high‑level proposals to detailed designs and ongoing maintenance.
• Lead and line‑manage a cross‑functional team (backend, frontend, DevOps, QA, documentation) across Europe and India.
• Design, build and operate a managed Slurm scheduling layer for bare‑metal GPU clusters.
• Develop and maintain the Kubernetes control plane, GPU operator integration, and day‑2 operations such as upgrades and recovery.
• Implement scalable managed inference services, including multi‑GPU/node parallelism and confidential‑compute capabilities.
• Establish observability (metrics, logging, alerting) and on‑call processes for the control plane and GPU fleet.
• Collaborate with partners, vendors and internal research teams to translate workload needs into platform specifications.
• Recruit and set technical standards for new engineers joining the platform team.

Required profile

• 8+ years of hands‑on engineering experience, with at least 3 years leading infrastructure teams.
• Proven experience operating Slurm at scale and managing bare‑metal GPU clusters.
• Deep knowledge of NVIDIA driver/CUDA lifecycle, InfiniBand/RDMA, and Linux kernel modules.
• Production‑grade Kubernetes expertise, including control‑plane operations, CNI/CSI, and custom operators.
• Experience with HPC storage solutions (VAST, Lustre, NFS) and large‑scale data movement.
• Strong observability background using Prometheus, Grafana, Loki and SLO management.
• Fluent in JavaScript/Node.js to review and guide control‑plane code.
• Excellent written and spoken English; comfortable working remotely across UTC‑0 to UTC+5:30.

Required skills

• Slurm
• NVIDIA driver, CUDA, MIG, Fabric Manager, NVSwitch
• InfiniBand, RDMA, SR‑IOV
• Linux kernel modules, PCIe passthrough, vfio‑pci, cgroups, namespaces
• Kubernetes (control plane, CNI, CSI, operators, custom controllers, multi‑tenancy)
• HPC storage (VAST, Lustre, NFS, NVMe caching)
• Observability tools (Prometheus, Grafana, Loki, SLOs, incident response)
• JavaScript, Node.js