Staff Operational Readiness Technical Program Manager, AI Infrastructure Deployment

.About Nscale and the Deployment PMO

Nscale builds and operates large-scale AI infrastructure, delivering GPU clusters for leading hyperscalers, AI labs and enterprises across North America, EMEA and APAC.

The Deployment PMO turns signed contracts into live AI infrastructure, coordinating the many teams and partners involved in bringing large GPU clusters online.

About the role

The Deployment TPM owns the last mile of every deployment: the point where a validated cluster becomes a service that Nscale operates and the customer runs on. You make sure nothing reaches the customer or the operations team half-built, undocumented or unsupported.

You track operational readiness for every cluster. That covers the core platform services that operations depends on, the handover package the customer signs, and the formal acceptance process through which the operations team takes the cluster into support. You turn dozens of owners, services and evidence folders into one clear answer: ready to hand off, or not yet, and why.

This is an individual contributor role in the Deployment PMO. Expect to travel up to 50% of the time, primarily to data center sites.

What you’ll do

Drive operational readiness

• Own the operational readiness checklist for each cluster, from network edge and management connectivity through to managed Kubernetes, with a named owner for every service. • Track bring-up in the right order: network edge and management, core services, cloud platform and storage, secrets management, remaining control planes, then application services. • Chase the services operations cannot run without: DHCP/PXE, DNS, NTP, zero-touch provisioning, LDAP, TACACS+, PKI, secrets management, asset systems, artifact and OS image mirrors, fleet management, logging, telemetry and observability. • Escalate early when a service or owner is blocking readiness, with evidence and a recovery date.

Build the handover package

• Assemble the customer handover document for every cluster: executive summary, scope, available GPU capacity against plan, metadata, roles and contacts, solution and network overview, and bill of materials. • Make capacity unambiguous: planned GPUs, GPUs available at handover, GPUs excluded and why, and open remediation tickets. • Map every validation test to its evidence folder (health checks, system and network stress, interconnect and multi-node performance, fabric and firmware validation) so the customer can verify results without asking. • Record every deviation, known issue and residual risk as an open item with an owner and a date. Nothing is accepted silently.