We’ll now “productionize” the Docker-based system from earlier using:

✅ Helm chart for easy installation and environment management

✅ Kubernetes manifests for API, worker, Redis, and Qdrant

✅ ConfigMaps / Secrets / Autoscaling setup

✅ PersistentVolumeClaim for Qdrant storage

Everything will be ready to helm install on any cluster (EKS, GKE, AKS, or local minikube).

---

⚙️ Folder structure

helm-ai-assistant/
├── Chart.yaml
├── values.yaml
├── templates/
│ ├── api-deployment.yaml
│ ├── worker-deployment.yaml
│ ├── redis-deployment.yaml
│ ├── qdrant-deployment.yaml
│ ├── api-service.yaml
│ ├── redis-service.yaml
│ ├── qdrant-service.yaml
│ ├── configmap.yaml
│ ├── secret.yaml
│ └── hpa.yaml
└── README.md

---

🧩 Chart.yaml

apiVersion: v2
name: ai-business-assistant
description: Helm chart for AI Business Assistant (RAG + Multi-Agent System)
type: application
version: 0.1.0
appVersion: "1.0"

---

⚙️ values.yaml

replicaCount:
api: 2
worker: 3

image:
api: "yourdockerhubuser/ai-assistant-api

"
worker: "yourdockerhubuser/ai-assistant-worker"
redis: "redis:7-alpine"
qdrant: "qdrant/qdrant.11.0"

service:
apiPort: 8000
redisPort: 6379
qdrantPort: 6333

resources:
api:
limits:
cpu: "500m"
memory: "512Mi"
worker:
limits:
cpu: "1"
memory: "1Gi"

env:
CELERY_BROKER_URL: "redis://redis:6379/0"
CELERY_RESULT_BACKEND: "redis://redis:6379/1"
QDRANT_URL: "http://qdrant:6333"
OPENAI_API_KEY: "changeme"

storage:
qdrant:
size: 1Gi

---

🗝️ templates/secret.yaml

apiVersion: v1
kind: Secret
metadata:
name: ai-secret
type: Opaque
stringData:
OPENAI_API_KEY: {{ .Values.env.OPENAI_API_KEY | quote }}

---

⚙️ templates/configmap.yaml

apiVersion: v1
kind: ConfigMap
metadata:
name: ai-config
data:
CELERY_BROKER_URL: {{ .Values.env.CELERY_BROKER_URL | quote }}
CELERY_RESULT_BACKEND: {{ .Values.env.CELERY_RESULT_BACKEND | quote }}
QDRANT_URL: {{ .Values.env.QDRANT_URL | quote }}

---

🚀 templates/api-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
name: ai-api
spec:
replicas: {{ .Values.replicaCount.api }}
selector:
matchLabels:
app: ai-api
template:
metadata:
labels:
app: ai-api
spec:
containers:
- name: api
image: {{ .Values.image.api }}
ports:
- containerPort: {{ .Values.service.apiPort }}
envFrom:
- configMapRef:
name: ai-config
- secretRef:
name: ai-secret
resources:
{{- toYaml .Values.resources.api | nindent 10 }}

apiVersion: v1
kind: Service
metadata:
name: ai-api
spec:
selector:
app: ai-api
ports:
- protocol: TCP
port: {{ .Values.service.apiPort }}
targetPort: {{ .Values.service.apiPort }}
type: LoadBalancer

---

⚙️ templates/worker-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
name: ai-worker
spec:
replicas: {{ .Values.replicaCount.worker }}
selector:
matchLabels:
app: ai-worker
template:
metadata:
labels:
app: ai-worker
spec:
containers:
- name: worker
image: {{ .Values.image.worker }}
command: ["celery", "-A", "tasks.celery", "worker", "--loglevel=info", "-Q", "retrieval,generation,critique,summarize,orch"]
envFrom:
- configMapRef:
name: ai-config
- secretRef:
name: ai-secret
resources:
{{- toYaml .Values.resources.worker | nindent 10 }}

---

⚙️ templates/redis-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
name: redis
spec:
replicas: 1
selector:
matchLabels:
app: redis
template:
metadata:
labels:
app: redis
spec:
containers:
- name: redis
image: {{ .Values.image.redis }}
ports:
- containerPort: {{ .Values.service.redisPort }}

apiVersion: v1
kind: Service
metadata:
name: redis
spec:
selector:
app: redis
ports:
- protocol: TCP
port: {{ .Values.service.redisPort }}
targetPort: {{ .Values.service.redisPort }}

---

⚙️ templates/qdrant-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
name: qdrant
spec:
replicas: 1
selector:
matchLabels:
app: qdrant
template:
metadata:
labels:
app: qdrant
spec:
containers:
- name: qdrant
image: {{ .Values.image.qdrant }}
ports:
- containerPort: {{ .Values.service.qdrantPort }}
volumeMounts:
- name: qdrant-storage
mountPath: /qdrant/storage
volumes:
- name: qdrant-storage
persistentVolumeClaim:
claimName: qdrant-pvc

apiVersion: v1
kind: Service
metadata:
name: qdrant
spec:
selector:
app: qdrant
ports:
- protocol: TCP
port: {{ .Values.service.qdrantPort }}
targetPort: {{ .Values.service.qdrantPort }}

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: qdrant-pvc
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: {{ .Values.storage.qdrant.size }}

---

⚙️ templates/hpa.yaml

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: ai-worker-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: ai-worker
minReplicas: 2
maxReplicas: 10
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 75

---

📘 README.md

AI Business Assistant Helm Chart

This chart deploys a production-ready RAG + Multi-Agent pipeline using Celery on Kubernetes.

🧭 Components

• API Gateway (FastAPI)
• Celery Workers
• Redis Broker
• Qdrant Vector Database
• Secrets & ConfigMaps
• Horizontal Pod Autoscaler

---

🚀 Quick Start

1️⃣ Prerequisites

• Kubernetes cluster (minikube, EKS, GKE)
• helm installed
• Docker images pushed (for api and worker)

2️⃣ Install the chart

helm install ai-assistant ./helm-ai-assistant --set env.OPENAI_API_KEY="sk-..."

3️⃣ Get service URL

kubectl get svc ai-api

Then open EXTERNAL-IP:8000 in your browser or call:

curl -X POST http://<EXTERNAL-IP>:8000/v1/query \
  -H "Content-Type: application/json" \
  -d '{"user_id":"u1","query":"AI trends in 2025"}'

4️⃣ Upgrade

helm upgrade ai-assistant ./helm-ai-assistant

5️⃣ Uninstall

helm uninstall ai-assistant


---

💡 Notes

Qdrant uses a PersistentVolume for vector data.

Scale workers automatically with the included HPA.

API is exposed as a LoadBalancer for external access.

Add ingress.yaml to route traffic through NGINX if required.


---

## 🧠 Next Step Options
Would you like me to:
1. **Generate ready-to-publish Dockerfiles** for `api` and `worker` (so you can push to Docker Hub for the Helm deployment)?  
2. Or directly **bundle this Helm chart as a `.tgz` package** you can upload to your cluster?  

Which one should I prepare next?