diff --git a/.github/workflows/helm.yaml b/.github/workflows/helm.yaml
new file mode 100644
index 000000000..f2b87af64
--- /dev/null
+++ b/.github/workflows/helm.yaml
@@ -0,0 +1,54 @@
+name: Publish Helm Chart
+
+on:
+  push:
+    branches:
+      - main
+      - dev
+    paths:
+      - "charts/**"
+env:
+  HELM_REGISTRY: ghcr.io
+  HELM_ORG: linagora
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Helm
+        uses: azure/setup-helm@v4
+        with:
+          version: v3.14.4
+
+      - name: Set chart version
+        run: |
+          BASE_VERSION=$(yq '.version' charts/openrag-stack/Chart.yaml)
+
+          if [[ "${GITHUB_REF_NAME}" == "main" ]]; then
+            # Keep version as-is for stable
+            CHART_VERSION="$BASE_VERSION"
+          else
+            # Append branch for dev builds
+            CHART_VERSION="${BASE_VERSION}-dev"
+            yq -i ".version = \"${CHART_VERSION}\"" charts/openrag-stack/Chart.yaml
+          fi
+
+          echo "CHART_VERSION=$CHART_VERSION" >> $GITHUB_ENV
+          echo "Using chart version: $CHART_VERSION"
+
+      - name: Helm dependency update
+        run: helm dependency update charts/openrag-stack
+
+      - name: Package chart
+        run: helm package charts/openrag-stack --destination .
+
+      - name: Login to GHCR
+        run: |
+          echo "${{ secrets.GITHUB_TOKEN }}" | helm registry login $HELM_REGISTRY -u ${{ github.actor }} --password-stdin
+
+      - name: Push chart
+        run: |
+          helm push openrag-stack-${CHART_VERSION}.tgz oci://$HELM_REGISTRY/$HELM_ORG
diff --git a/.gitignore b/.gitignore
index 1d15c51a1..ad803d074 100644
--- a/.gitignore
+++ b/.gitignore
@@ -67,3 +67,6 @@ services/*
 *.json
 *.csv
 *.pkl
+
+#helm
+charts/openrag-stack/charts/*.tgz
diff --git a/Dockerfile.ray b/Dockerfile.ray
index 31e87d385..8f91d86c7 100644
--- a/Dockerfile.ray
+++ b/Dockerfile.ray
@@ -12,6 +12,7 @@ RUN apt-get update && apt-get install -y \
     make \
     ssh \ 
     rsync \
+    wget \
     libpq-dev python3-dev \
     && rm -rf /var/lib/apt/lists/*
 
@@ -35,7 +36,7 @@ COPY pyproject.toml uv.lock ./
 RUN pip3 install uv && \
     uv python install 3.12.7 && \
     uv python pin 3.12.7 
-    # && \uv sync --no-dev
+# && \uv sync --no-dev
 COPY entrypoint.sh /app/entrypoint.sh
 RUN chmod +x /app/entrypoint.sh
 
@@ -46,10 +47,10 @@ WORKDIR /app/openrag
 COPY openrag/ .
 
 # Copy assests & config
-COPY public/ /app/public/
 COPY prompts/ /app/prompts/
 COPY .hydra_config/ /app/.hydra_config/
 
 RUN apt install -y 
+RUN ln -s /app/.venv/bin/ray /usr/local/bin/ray
 
 ENV PYTHONPATH=/app/openrag/
\ No newline at end of file
diff --git a/charts/openrag-stack/.helmignore b/charts/openrag-stack/.helmignore
new file mode 100644
index 000000000..0e8a0eb36
--- /dev/null
+++ b/charts/openrag-stack/.helmignore
@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/
diff --git a/charts/openrag-stack/Chart.lock b/charts/openrag-stack/Chart.lock
new file mode 100644
index 000000000..9001e5e43
--- /dev/null
+++ b/charts/openrag-stack/Chart.lock
@@ -0,0 +1,15 @@
+dependencies:
+- name: kuberay-operator
+  repository: https://ray-project.github.io/kuberay-helm/
+  version: 1.4.0
+- name: postgresql
+  repository: https://charts.bitnami.com/bitnami
+  version: 15.5.23
+- name: milvus
+  repository: https://zilliztech.github.io/milvus-helm/
+  version: 5.0.0
+- name: vllm-stack
+  repository: https://vllm-project.github.io/production-stack
+  version: 0.1.7
+digest: sha256:eb9f465f19be0d6d6c563c4eda6dc0f7b8e9331209420f14f01587f2ca29f965
+generated: "2025-09-04T13:28:20.5556128Z"
diff --git a/charts/openrag-stack/Chart.yaml b/charts/openrag-stack/Chart.yaml
new file mode 100644
index 000000000..cdf3906cf
--- /dev/null
+++ b/charts/openrag-stack/Chart.yaml
@@ -0,0 +1,39 @@
+apiVersion: v2
+name: openrag-stack
+description: A Helm chart for Kubernetes
+
+# A chart can be either an 'application' or a 'library' chart.
+#
+# Application charts are a collection of templates that can be packaged into versioned archives
+# to be deployed.
+#
+# Library charts provide useful utilities or functions for the chart developer. They're included as
+# a dependency of application charts to inject those utilities and functions into the rendering
+# pipeline. Library charts do not define any templates and therefore cannot be deployed.
+type: application
+
+# This is the chart version. This version number should be incremented each time you make changes
+# to the chart and its templates, including the app version.
+# Versions are expected to follow Semantic Versioning (https://semver.org/)
+version: 0.2.0
+
+# This is the version number of the application being deployed. This version number should be
+# incremented each time you make changes to the application. Versions are not expected to
+# follow Semantic Versioning. They should reflect the version the application is using.
+# It is recommended to use it with quotes.
+appVersion: "1.2.0"
+
+dependencies:
+  - name: kuberay-operator
+    version: "1.4.0"
+    repository: "https://ray-project.github.io/kuberay-helm/"
+  - name: postgresql
+    version: "15.5.23"
+    repository: "https://charts.bitnami.com/bitnami"
+  - name: milvus
+    version: "5.0.0"
+    repository: "https://zilliztech.github.io/milvus-helm/"
+  - name: vllm-stack
+    alias: vllm
+    version: ">=0.1.7"
+    repository: "https://vllm-project.github.io/production-stack"
diff --git a/charts/openrag-stack/templates/_helpers.tpl b/charts/openrag-stack/templates/_helpers.tpl
new file mode 100644
index 000000000..c5bd0ede7
--- /dev/null
+++ b/charts/openrag-stack/templates/_helpers.tpl
@@ -0,0 +1,62 @@
+{{/*
+Expand the name of the chart.
+*/}}
+{{- define "openrag-stack.name" -}}
+{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Create a default fully qualified app name.
+We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec).
+If release name contains chart name it will be used as a full name.
+*/}}
+{{- define "openrag-stack.fullname" -}}
+{{- if .Values.fullnameOverride }}
+{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- $name := default .Chart.Name .Values.nameOverride }}
+{{- if contains $name .Release.Name }}
+{{- .Release.Name | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" }}
+{{- end }}
+{{- end }}
+{{- end }}
+
+{{/*
+Create chart name and version as used by the chart label.
+*/}}
+{{- define "openrag-stack.chart" -}}
+{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Common labels
+*/}}
+{{- define "openrag-stack.labels" -}}
+helm.sh/chart: {{ include "openrag-stack.chart" . }}
+{{ include "openrag-stack.selectorLabels" . }}
+{{- if .Chart.AppVersion }}
+app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
+{{- end }}
+app.kubernetes.io/managed-by: {{ .Release.Service }}
+{{- end }}
+
+{{/*
+Selector labels
+*/}}
+{{- define "openrag-stack.selectorLabels" -}}
+app.kubernetes.io/name: {{ include "openrag-stack.name" . }}
+app.kubernetes.io/instance: {{ .Release.Name }}
+{{- end }}
+
+{{/*
+Create the name of the service account to use
+*/}}
+{{- define "openrag-stack.serviceAccountName" -}}
+{{- if .Values.serviceAccount.create }}
+{{- default (include "openrag-stack.fullname" .) .Values.serviceAccount.name }}
+{{- else }}
+{{- default "default" .Values.serviceAccount.name }}
+{{- end }}
+{{- end }}
diff --git a/charts/openrag-stack/templates/configmap-env.yaml b/charts/openrag-stack/templates/configmap-env.yaml
new file mode 100644
index 000000000..fa1a5da9e
--- /dev/null
+++ b/charts/openrag-stack/templates/configmap-env.yaml
@@ -0,0 +1,8 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: rag-env
+data:
+{{- range $key, $value := .Values.env.config }}
+  {{ $key }}: "{{ tpl (printf "%v" $value) $ }}"
+{{- end }}
diff --git a/charts/openrag-stack/templates/infinity.yaml b/charts/openrag-stack/templates/infinity.yaml
new file mode 100644
index 000000000..cafcfd51f
--- /dev/null
+++ b/charts/openrag-stack/templates/infinity.yaml
@@ -0,0 +1,60 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: {{ .Release.Name }}-reranker
+  labels:
+    app.kubernetes.io/name: reranker
+    app.kubernetes.io/instance: {{ .Release.Name }}
+spec:
+  replicas: {{ .Values.reranker.replicas }}
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: reranker
+      app.kubernetes.io/instance: {{ .Release.Name }}
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/name: reranker
+        app.kubernetes.io/instance: {{ .Release.Name }}
+    spec:
+      runtimeClassName: {{ .Values.reranker.runtimeClassName }}
+      containers:
+        - name: reranker
+          image: "{{ .Values.reranker.image.repository }}:{{ .Values.reranker.image.tag }}"
+          args:
+            - "v2"
+            - "--model-id"
+            - "{{ .Values.reranker.model.id }}"
+            - "--port"
+            - "{{ .Values.reranker.service.port }}"
+          ports:
+            - containerPort: {{ .Values.reranker.service.port }}
+          resources:
+            {{- toYaml .Values.reranker.resources | nindent 12 }}
+          volumeMounts:
+            - name: hf-cache
+              mountPath: /app/.cache/huggingface
+          envFrom:
+            - configMapRef:
+                name: rag-env
+      volumes:
+        - name: hf-cache
+          persistentVolumeClaim:
+            claimName: rag-model-weights
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ .Release.Name }}-reranker
+  labels:
+    app.kubernetes.io/name: reranker
+    app.kubernetes.io/instance: {{ .Release.Name }}
+spec:
+  selector:
+    app.kubernetes.io/name: reranker
+    app.kubernetes.io/instance: {{ .Release.Name }}
+  ports:
+    - protocol: TCP
+      port: {{ .Values.reranker.service.port }}
+      targetPort: {{ .Values.reranker.service.port }}
+  type: {{ .Values.reranker.service.type }}
diff --git a/charts/openrag-stack/templates/ingress.yaml b/charts/openrag-stack/templates/ingress.yaml
new file mode 100644
index 000000000..255e16634
--- /dev/null
+++ b/charts/openrag-stack/templates/ingress.yaml
@@ -0,0 +1,26 @@
+{{- if .Values.ingress.enabled }}
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: {{ include "openrag-stack.fullname" . }}-ingress
+  annotations:
+    nginx.ingress.kubernetes.io/rewrite-target: /$1
+    nginx.ingress.kubernetes.io/use-regex: "true"  
+spec:
+  ingressClassName: {{ .Values.ingress.className | quote }}
+  rules:
+    - http:
+        paths:
+          {{- if .Values.ingress.paths.openrag.enabled }}
+          - path: /(.*)
+            pathType: ImplementationSpecific
+            backend:
+              service:
+                name: raycluster-head-svc
+                port:
+                  number: {{ .Values.env.config.RAY_SERVE_PORT }}
+          {{- end }}
+      {{- with .Values.ingress.host }}
+      host: {{ . | quote }}
+      {{- end }}
+{{- end }}
diff --git a/charts/openrag-stack/templates/openrag.yaml b/charts/openrag-stack/templates/openrag.yaml
new file mode 100644
index 000000000..e97a0e67e
--- /dev/null
+++ b/charts/openrag-stack/templates/openrag.yaml
@@ -0,0 +1,85 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: {{ .Release.Name }}-openrag
+  labels:
+    app.kubernetes.io/name: openrag
+    app.kubernetes.io/instance: {{ .Release.Name }}
+spec:
+  replicas: {{ .Values.openrag.replicas }}
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: openrag
+      app.kubernetes.io/instance: {{ .Release.Name }}
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/name: openrag
+        app.kubernetes.io/instance: {{ .Release.Name }}
+    spec:
+      initContainers:
+        - name: init-venv
+          image: "{{ .Values.openrag.image.repository }}:{{ .Values.openrag.image.tag }}"
+          command:
+            - sh
+            - -c
+            - |
+              echo "Waiting for Python env in /app/.venv..."
+              while [ ! -f /app/.venv/.ready ]; do
+                sleep 3
+              done
+              echo "Venv is ready."
+          volumeMounts:
+            - name: venv
+              mountPath: /app/.venv
+      containers:
+        - name: openrag
+          image: "{{ .Values.openrag.image.repository }}:{{ .Values.openrag.image.tag }}"
+          ports:
+            - containerPort: {{ .Values.openrag.service.port }}
+          resources:
+            {{- toYaml .Values.openrag.resources | nindent 12 }}
+          envFrom:
+            - configMapRef:
+                name: rag-env
+            - secretRef:
+                name: rag-env-secrets
+          volumeMounts:
+            - name: model-weights
+              mountPath: /app/model_weights
+            - name: data
+              mountPath: /app/data
+            - name: logs
+              mountPath: /app/logs
+            - name: venv
+              mountPath: /app/.venv
+      volumes:
+        - name: model-weights
+          persistentVolumeClaim:
+            claimName: rag-model-weights
+        - name: data
+          persistentVolumeClaim:
+            claimName: rag-data
+        - name: logs
+          persistentVolumeClaim:
+            claimName: rag-logs
+        - name: venv
+          persistentVolumeClaim:
+            claimName: rag-venv
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ .Release.Name }}-openrag
+  labels:
+    app.kubernetes.io/name: openrag
+    app.kubernetes.io/instance: {{ .Release.Name }}
+spec:
+  selector:
+    app.kubernetes.io/name: openrag
+    app.kubernetes.io/instance: {{ .Release.Name }}
+  ports:
+    - protocol: TCP
+      port: {{ .Values.openrag.service.port }}
+      targetPort: {{ .Values.openrag.service.port }}
+  type: {{ .Values.openrag.service.type }}
diff --git a/charts/openrag-stack/templates/pvc.yaml b/charts/openrag-stack/templates/pvc.yaml
new file mode 100644
index 000000000..050fdfc5c
--- /dev/null
+++ b/charts/openrag-stack/templates/pvc.yaml
@@ -0,0 +1,49 @@
+{{- if .Values.persistence.enabled }}
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: rag-model-weights
+spec:
+  accessModes:
+    - {{ .Values.persistence.accessMode }}
+  resources:
+    requests:
+      storage: {{ .Values.persistence.volumes.modelWeights.size }}
+  storageClassName: {{ .Values.persistence.storageClass }}
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: rag-data
+spec:
+  accessModes:
+    - {{ .Values.persistence.accessMode }}
+  resources:
+    requests:
+      storage: {{ .Values.persistence.volumes.data.size }}
+  storageClassName: {{ .Values.persistence.storageClass }}
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: rag-logs
+spec:
+  accessModes:
+    - {{ .Values.persistence.accessMode }}
+  resources:
+    requests:
+      storage: {{ .Values.persistence.volumes.logs.size }}
+  storageClassName: {{ .Values.persistence.storageClass }}
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: rag-venv
+spec:
+  accessModes:
+    - {{ .Values.persistence.accessMode }}
+  resources:
+    requests:
+      storage: {{ .Values.persistence.volumes.venv.size }}
+  storageClassName: {{ .Values.persistence.storageClass }}
+{{- end }}
diff --git a/charts/openrag-stack/templates/raycluster.yaml b/charts/openrag-stack/templates/raycluster.yaml
new file mode 100644
index 000000000..2c8408c1e
--- /dev/null
+++ b/charts/openrag-stack/templates/raycluster.yaml
@@ -0,0 +1,148 @@
+apiVersion: ray.io/v1
+kind: RayCluster
+metadata:
+  name: raycluster
+  annotations:
+    ray.io/overwrite-container-cmd: "true"
+spec:
+  rayVersion: "2.47.1"
+  enableInTreeAutoscaling: false
+
+  headGroupSpec:
+    serviceType: ClusterIP
+    template:
+      spec:
+        initContainers:
+          - name: init-venv-sync
+            image: {{ .Values.ray.image.repository }}:{{ .Values.ray.image.tag }}
+            command:
+              - sh
+              - -c
+              - |
+                if [ -f /app/.venv/.ready ]; then
+                  echo "Existing env detected, skipping install."
+                else
+                  echo "No env ready, cleaning and syncing..."
+                  rm -rf /app/.venv/*
+                  uv sync
+                  touch /app/.venv/.ready
+                fi
+            volumeMounts:
+              - name: venv
+                mountPath: /app/.venv
+
+        containers:
+          - name: ray-head
+            image: {{ .Values.ray.image.repository }}:{{ .Values.ray.image.tag }}
+            command: ["uv"]
+            args:
+              - "run"
+              - "ray"
+              - "start"
+              - "--head"
+              - "--dashboard-host=0.0.0.0"
+              - "--dashboard-agent-listen-port=52365"
+              - "--metrics-export-port=8080"
+              - "--block"
+            ports:
+              - containerPort: 80
+                name: serve
+              - containerPort: 8265
+                name: dashboard
+              - containerPort: 10001
+                name: client
+              - containerPort: 6379
+                name: gcs
+              - containerPort: 8080
+                name: metrics
+            volumeMounts:
+              - name: model-weights
+                mountPath: /app/model_weights
+              - name: data
+                mountPath: /app/data
+              - name: logs
+                mountPath: /app/logs
+              - name: venv
+                mountPath: /app/.venv
+            envFrom:
+              - configMapRef:
+                  name: rag-env
+              - secretRef:
+                  name: rag-env-secrets
+            resources:
+              limits:
+                nvidia.com/gpu: 1
+        volumes:
+          - name: model-weights
+            persistentVolumeClaim:
+              claimName: rag-model-weights
+          - name: data
+            persistentVolumeClaim:
+              claimName: rag-data
+          - name: logs
+            persistentVolumeClaim:
+              claimName: rag-logs
+          - name: venv
+            persistentVolumeClaim:
+              claimName: rag-venv
+
+  workerGroupSpecs:
+  {{- range $i, $e := until (.Values.ray.workers.count | int) }}
+    - groupName: worker-group-{{ add $i 1 }}
+      replicas: {{ $.Values.ray.workers.replicas }}
+      minReplicas: {{ $.Values.ray.workers.minReplicas }}
+      maxReplicas: {{ $.Values.ray.workers.maxReplicas }}
+      template:
+        spec:
+          initContainers:
+            - name: init-venv
+              image: ghcr.io/linagora/openrag:dev-latest
+              command:
+                - sh
+                - -c
+                - |
+                  echo "Waiting for Python env in /app/.venv..."
+                  while [ ! -f /app/.venv/.ready ]; do
+                    sleep 3
+                  done
+                  echo "Venv is ready."
+              volumeMounts:
+                - name: venv
+                  mountPath: /app/.venv
+          containers:
+            - name: ray-worker
+              image: {{ $.Values.ray.image.repository }}:{{ $.Values.ray.image.tag }}
+              command: ["/bin/bash", "-lc", "--"]
+              args:
+                - uv run $KUBERAY_GEN_RAY_START_CMD
+              volumeMounts:
+                - name: model-weights
+                  mountPath: /app/model_weights
+                - name: data
+                  mountPath: /app/data
+                - name: logs
+                  mountPath: /app/logs
+                - name: venv
+                  mountPath: /app/.venv
+              envFrom:
+                - configMapRef:
+                    name: rag-env
+                - secretRef:
+                    name: rag-env-secrets
+              resources:
+                limits:
+                  nvidia.com/gpu: 1
+          volumes:
+            - name: model-weights
+              persistentVolumeClaim:
+                claimName: rag-model-weights
+            - name: data
+              persistentVolumeClaim:
+                claimName: rag-data
+            - name: logs
+              persistentVolumeClaim:
+                claimName: rag-logs
+            - name: venv
+              persistentVolumeClaim:
+                claimName: rag-venv
+  {{- end }}
diff --git a/charts/openrag-stack/templates/secrets-env.yaml b/charts/openrag-stack/templates/secrets-env.yaml
new file mode 100644
index 000000000..5bf3ee1b8
--- /dev/null
+++ b/charts/openrag-stack/templates/secrets-env.yaml
@@ -0,0 +1,9 @@
+apiVersion: v1
+kind: Secret
+metadata:
+  name: rag-env-secrets
+type: Opaque
+stringData:
+{{- range $key, $value := .Values.env.secrets }}
+  {{ $key }}: "{{ $value }}"
+{{- end }}
diff --git a/charts/openrag-stack/values.yaml b/charts/openrag-stack/values.yaml
new file mode 100644
index 000000000..50d351023
--- /dev/null
+++ b/charts/openrag-stack/values.yaml
@@ -0,0 +1,224 @@
+# === Global shared persistence ===
+persistence:
+  enabled: true
+  storageClass: &storageClass longhorn
+  accessMode: ReadWriteMany
+  volumes:
+    modelWeights: { size: 50Gi }
+    data: { size: 20Gi }
+    hydraConfig: { size: 10Mi }
+    logs: { size: 1Gi }
+    venv: { size: 10Gi }
+
+ray:
+  workers:
+    count: 2
+    replicas: 1
+    minReplicas: 1
+    maxReplicas: 1
+  image:
+    repository: ghcr.io/linagora/openrag-ray
+    tag: latest
+
+# === PostgreSQL (bitnami) ===
+postgresql:
+  enabled: true
+  auth:
+    username: &pgUser root
+    password: &pgPass root_password
+  primary:
+    persistence:
+      enabled: true
+      size: 10Gi
+      storageClass: *storageClass
+  service:
+    port: &pgPort 5432
+
+# === Milvus ===
+milvus:
+  enabled: true
+  image:
+    all: { tag: v2.6.0 }
+  cluster: { enabled: true }
+  pulsarv3: { enabled: false }
+  woodpecker: { enabled: true }
+  streaming: { enabled: true }
+  indexNode: { enabled: false }
+  minio:
+    mode: distributed
+    replicas: 4
+    persistence:
+      enabled: true
+      size: 20Gi
+      storageClass: *storageClass
+  etcd:
+    replicaCount: 3
+    persistence:
+      enabled: true
+      size: 10Gi
+      storageClass: *storageClass
+  proxy:
+    service:
+      type: ClusterIP
+      port: 19530
+
+# === vLLM (embedder) ===
+vllm:
+  embedderModelName: &embedderModel "Qwen/Qwen3-Embedding-0.6B"
+  servingEngineSpec:
+    enableEngine: true
+    modelSpec:
+      - name: "embedder"
+        repository: "vllm/vllm-openai"
+        tag: "latest"
+        modelURL: *embedderModel
+        replicaCount: 1
+        requestCPU: 4
+        requestMemory: "16Gi"
+        requestGPU: 1
+        limitCPU: 4
+        limitMemory: "16Gi"
+        hf_token:
+          secretName: rag-env-secrets
+          secretKey: HF_TOKEN
+        vllmConfig:
+          gpuMemoryUtilization: 0.3
+          extraArgs: ["--trust-remote-code", "--task", "embed"]
+    servicePort: &vllmPort 8000
+
+# === Infinity reranker ===
+reranker:
+  rerankerModelName: &rerankerModel "Alibaba-NLP/gte-multilingual-reranker-base"
+  servicePort: &rerankerPort 7997
+  image:
+    repository: michaelf34/infinity
+    tag: latest
+  model:
+    id: Alibaba-NLP/gte-multilingual-reranker-base
+  command:
+    - v2
+    - --model-id
+    - *rerankerModel
+    - --port
+    - *rerankerPort
+  replicas: 1
+  runtimeClassName: nvidia
+  service:
+    type: ClusterIP
+    port: *rerankerPort
+  resources:
+    requests:
+      cpu: "4"
+      memory: "16Gi"
+    limits:
+      cpu: "4"
+      memory: "16Gi"
+# === OpenRAG main app ===
+openrag:
+  image:
+    repository: ghcr.io/linagora/openrag
+    tag: latest
+  service:
+    type: ClusterIP
+    port: 8080
+  resources:
+    requests:
+      cpu: "4"
+      memory: "16Gi"
+    limits:
+      cpu: "4"
+      memory: "16Gi"
+  replicas: 1
+
+ingress:
+  enabled: true
+  className: "nginx"
+  host: ""
+
+  paths:
+    openrag:
+      enabled: true
+
+# === Shared env (config + secrets) ===
+env:
+  config:
+    # LLM
+    BASE_URL: "https://example.com/v1/"
+    MODEL: "Mistral-Small-3.1-24B-Instruct-2503"
+    LLM_SEMAPHORE: "50"
+
+    # VLM
+    VLM_BASE_URL: "https://example.com/v1/"
+    VLM_MODEL: "Qwen2.5-VL-7B-Instruct"
+    VLM_SEMAPHORE: "50"
+
+    # App
+    APP_PORT: "8080"
+    APP_HOST: "0.0.0.0"
+    ENABLE_RAY_SERVE: "true"
+    RAY_SERVE_NUM_REPLICAS: "4"
+    RAY_SERVE_PORT: "80"
+    DISABLE_EXCEPTION_HANDLER: "true"
+
+    WITH_CHAINLIT_UI: "false"
+    SAVE_UPLOADED_FILES: "false"
+    API_NUM_WORKERS: "8"
+    INDEXERUI_URL: "http://indexer-ui:3042"
+
+    # Vector DB
+    VDB_HOST: "{{ .Release.Name }}-milvus"
+    VDB_PORT: "19530"
+    VDB_CONNECTOR_NAME: "milvus"
+
+    # PostgreSQL
+    POSTGRES_HOST: "{{ .Release.Name }}-postgresql"
+    POSTGRES_PORT: *pgPort
+    POSTGRES_USER: *pgUser
+    POSTGRES_PASSWORD: *pgPass
+
+    # Retriever
+    CONTEXTUAL_RETRIEVAL: "true"
+    RETRIEVER_TOP_K: "15"
+
+    # Embedder
+    EMBEDDER_MODEL_NAME: *embedderModel
+    EMBEDDER_BASE_URL: "http://{{ .Release.Name }}-embedder-engine-service:{{ .Values.vllm.servingEngineSpec.servicePort }}/v1"
+
+    # Reranker
+    RERANKER_MODEL: *rerankerModel
+    RERANKER_ENABLED: "true"
+    RERANKER_TOP_K: "4"
+    RERANKER_BASE_URL: "http://{{ .Release.Name }}-reranker:{{ .Values.reranker.servicePort }}"
+    RERANKER_MODEL_TYPE: "infinity"
+
+    # Prompts
+    PROMPTS_DIR: "../prompts/example3"
+    COLBERT_LOAD_TORCH_EXTENSION_VERBOSE: "True"
+
+    # Loaders
+    PDFLoader: "MarkerLoader"
+    XDG_CACHE_HOME: "/app/model_weights"
+    MARKER_MAX_TASKS_PER_CHILD: "10"
+    MARKER_MAX_PROCESSES: "15"
+    MARKER_MIN_PROCESSES: "3"
+    MARKER_POOL_SIZE: "3"
+    MARKER_NUM_GPUS: "0.6"
+
+    # Ray
+    RAY_NUM_GPUS: "0.1"
+    RAY_POOL_SIZE: "3"
+    RAY_MAX_TASKS_PER_WORKER: "50"
+    RAY_DASHBOARD_PORT: "8265"
+    RAY_ADDRESS: "ray://raycluster-head-svc:10001"
+    RAY_task_retry_delay_ms: "3000"
+    RAY_ENABLE_UV_RUN_RUNTIME_ENV: "0"
+
+    UV_LINK_MODE: "copy"
+    UV_CACHE_DIR: "/tmp/uv-cache"
+
+  secrets:
+    API_KEY: "sk-xxxx" # LLM API KEY
+    VLM_API_KEY: "sk-xxxx" # VLM API KEY
+    EMBEDDER_API_KEY: "EMPTY"
+    AUTH_TOKEN: "sk-xxxx" # API KEY for OpenRAG
+    HF_TOKEN: "hf_xxxx" # HuggingFace token
diff --git a/docs/kubernetes.md b/docs/kubernetes.md
new file mode 100644
index 000000000..8d045e788
--- /dev/null
+++ b/docs/kubernetes.md
@@ -0,0 +1,51 @@
+# Deploying OpenRAG on Kubernetes
+
+This guide explains how to deploy the **OpenRAG** stack on a Kubernetes cluster using Helm.
+
+---
+
+## Prerequisites
+
+- A **Kubernetes cluster** with **GPU nodes** available (NVIDIA runtime) and nvidia-gpu-operator installed.
+- A **StorageClass** that supports **ReadWriteMany** (`RWX`) access mode.  
+  This is required because the Ray cluster workers and the OpenRAG app need to access the same shared volumes (e.g. for `.venv`, model weights, logs, data).
+- If using ingress, the ingress-nginx controller needs to be installed on the cluster.
+
+---
+
+## Steps
+
+1. **Create a `values.yaml` file**:
+
+   - Copy or create a new `values.yaml` at the root of your repo.
+   - You can see the full example file inside the chart:
+     [../charts/openrag-stack/values.yaml](../charts/openrag-stack/values.yaml)
+   - Customize the values you need (e.g., image tags, resources, ingress host, storage class, environment variables, secrets).
+
+2. **Set environment and secrets**:
+
+   - Edit the `env.config` and `env.secrets` sections in your `values.yaml`.
+   - Secrets (API keys, tokens, Hugging Face credentials, etc.) will be mounted into the cluster as Kubernetes secrets.
+
+3. **Install or upgrade the release from GHCR**:
+
+   ```bash
+   helm upgrade --install openrag oci://ghcr.io/linagora/openrag-stack -f ./values.yaml --version 0.1.0
+   ```
+
+   - `openrag` is the Helm release name.
+   - `oci://ghcr.io/linagora/openrag-stack` is the remote chart location.
+   - `-f ./values.yaml` specifies your custom configuration.
+   - `--version 0.1.0` ensures you deploy a specific chart version.
+
+---
+
+## Notes
+
+- If using a public IP instead of a hostname, you can leave `ingress.host` empty in your `values.yaml`.  
+  The ingress will then match all hosts.
+
+- If you later configure a hostname + TLS (via cert-manager), just update `ingress.host` and redeploy.
+
+- Ensure your GPU nodes have the correct NVIDIA drivers and `nvidia` `RuntimeClass` configured.
+
diff --git a/openrag/api.py b/openrag/api.py
index 1b3029620..5f9552519 100644
--- a/openrag/api.py
+++ b/openrag/api.py
@@ -63,6 +63,9 @@ def __init__(self, config):
 INDEXERUI_COMPOSE_FILE = os.getenv("INDEXERUI_COMPOSE_FILE", None)
 INDEXERUI_PORT: Optional[str] = os.getenv("INDEXERUI_PORT", "3042")
 
+DISABLE_EXCEPTION_HANDLER: bool = (
+    os.getenv("DISABLE_EXCEPTION_HANDLER", "false").lower() == "true"
+)
 
 security = HTTPBearer()
 
@@ -84,10 +87,12 @@ def verify_token(credentials: HTTPAuthorizationCredentials = Depends(security)):
 
 
 # Exception handlers
-@app.exception_handler(OpenRAGError)
-async def openrag_exception_handler(request: Request, exc: OpenRAGError):
-    logger.error("OpenRAGError occurred", error=str(exc))
-    return JSONResponse(status_code=exc.status_code, content=exc.to_dict())
+if not DISABLE_EXCEPTION_HANDLER:
+
+    @app.exception_handler(OpenRAGError)
+    async def openrag_exception_handler(request: Request, exc: OpenRAGError):
+        logger.error("OpenRAGError occurred", error=str(exc))
+        return JSONResponse(status_code=exc.status_code, content=exc.to_dict())
 
 
 # Add CORS middleware
diff --git a/openrag/components/indexer/loaders/pdf_loaders/marker.py b/openrag/components/indexer/loaders/pdf_loaders/marker.py
index 1ad281451..b6b1555f4 100644
--- a/openrag/components/indexer/loaders/pdf_loaders/marker.py
+++ b/openrag/components/indexer/loaders/pdf_loaders/marker.py
@@ -160,7 +160,7 @@ def __init__(self):
         self.actors = [MarkerWorker.remote() for _ in range(self.pool_size)]
         self._queue: asyncio.Queue[ray.actor.ActorHandle] = asyncio.Queue()
 
-        for _ in range(self.pool_size):
+        for _ in range(self.max_processes):
             for actor in self.actors:
                 self._queue.put_nowait(actor)