Why should you care?

Bulk reindex of hospital PDFs on CPU takes days; GPU embedding servers cut nightly jobs to hours.

See it live — copy this example

Paste into an ASP.NET Core 8+ / AgentNest project, then run with dotnet run (set your API keys in user-secrets).

# agentnest-embed-gpu-job.yaml
apiVersion: batch/v1
kind: Job
metadata:
  name: embed-reindex-tenant-42
spec:
  template:
    spec:
      containers:
      - name: embedder
        image: agentnest/embed-worker:1.4.0
        resources:
          limits:
            nvidia.com/gpu: 1
        env:
        - name: TENANT_ID
          value: "42"
      restartPolicy: Never
      nodeSelector:
        accelerator: nvidia-a10

What happened?

• Kubernetes Job requests one GPU on A10 nodes to run embed-worker for tenant 42 reindex without blocking API pods.
• Follow the steps below — typing the code yourself is the fastest way to learn.

Practice next

1. Enable NVIDIA device plugin on AKS GPU node pool.
2. Build embed-worker container calling local or vLLM endpoint.
3. Queue reindex jobs from AgentNest admin API.
4. Switch Job to CronJob for nightly all-tenant reindex.
5. Share GPU via time-slicing for small dev clusters.

Remember

Use K8s Jobs with nvidia.com/gpu for batch embedding. Isolate GPU workloads from interactive API pods. Target GPU nodes with nodeSelector labels.

Hospital corpus reindex

10GB protocol update must embed before Monday go-live.

Outcome: GPU Job finishes reindex in 3 hours; Qdrant serves new vectors for triage agent.