Helm Chart

Added in v3.11.0

Deploy Director-AI to Kubernetes with the bundled Helm chart at deploy/helm/director-ai/.

Install

helm install director-ai deploy/helm/director-ai/ \
  --set server.profile=thorough \
  --set server.apiKeySecret=director-api-key

Configuration

Key values in values.yaml:

Value	Default	Description
replicaCount	1	Number of replicas
image.repository	ghcr.io/anulum/director-ai	Container image
server.port	8080	API server port
server.workers	2	Uvicorn workers
server.profile	fast	Scoring profile (fast, thorough, medical, etc.)
server.apiKeySecret	""	K8s secret name containing api-key
gpu.enabled	false	Enable GPU resources
autoscaling.enabled	false	Enable HPA
autoscaling.targetCPUUtilization	70	HPA CPU target

GPU Mode

helm install director-ai deploy/helm/director-ai/ \
  --set gpu.enabled=true \
  --set server.profile=thorough

API Key

Create a Kubernetes secret first:

kubectl create secret generic director-api-key \
  --from-literal=api-key=your-secret-key

Then reference it:

helm install director-ai deploy/helm/director-ai/ \
  --set server.apiKeySecret=director-api-key

Security

The chart enforces:

• runAsNonRoot: true (UID 1000)
• readOnlyRootFilesystem: true
• All capabilities dropped
• allowPrivilegeEscalation: false

Health Probes

• Liveness: GET /health (initial delay 10s, period 30s)
• Readiness: GET /health (initial delay 5s, period 10s)