AI Infrastructure Ops Lead - GPU, Kubernetes & Performance
Posted Updated
Description
Malomatia seeks a Consultant for Operations Oversight & Performance, AI Infrastructure to oversee post-go-live delivery of GPU-based AI runtimes on Kubernetes, ensuring performance, reliability, and continuous improvement.
You will define KPIs with third parties, harden clusters with GPU operators, review scheduling, manage AI workloads, and drive observability and GitOps across multi-cluster environments.
Define, agree, and monitor operational KPIs and SLAs with 3rd party (availability, incident response, resolution times, change success rate, capacity utilization). Escort Operation and hardening of Kubernetes clusters with GPU Operator, device plugins, DRA, MIG/time-slicing, and advanced schedulers (Kueue or equivalents). Review GPU-aware scheduling, fair-share/quotas, priority classes, preemption, and multi-tenancy isolation. Supervise manages container runtime, AI workloads (training/inference), and autoscaling. Review observability (Prometheus/Grafana + DCGM), logging, and GitOps for the runtime layer. Support multi-cluster federation and cluster lifecycle (in partnership with Rafay layer). Conduct regular operational reviews of Advanced NOC performance, incident trends, problem management, and change records. Drive service improvement plans and hold 3rd party accountable for operational performance. Review capacity, lifecycle, and maintenance plans; escalat risks to the Program Delivery Lead. Ensure operational processes support future Platform foundation requirements (observability, identity, scheduling readiness). Coordinate planned works, maintenance windows, and operational communications with internal stakeholders. Maintain the operational risk register and support audit/compliance evidence collection. 10 years in GPU HPC infrastructure operations, service management, or managed-service oversight roles. 7+ years Kubernetes (CKA preferred); deep experience running GPU workloads at scale. Strong knowledge of CNCF ecosystem, operators, CRDs, and GPU-specific features (NVIDIA GPU Operator, DRA). Excellent understanding of GPU or high-performance compute operations, observability, and incident/change management. Strong analytical, reporting, and stakeholder management skills.Skills
Sign in to read and join the discussion.