diff --git a/deployments/brev/BUILD b/deployments/brev/BUILD
index 37eb52b3c..c02f30604 100644
--- a/deployments/brev/BUILD
+++ b/deployments/brev/BUILD
@@ -37,7 +37,7 @@ config_setting(
 sh_test(
     name = "shellcheck",
     srcs = ["shellcheck_test.sh"],
-    data = ["setup.sh"] + select({
+    data = ["setup.sh", "setup-k8s.sh"] + select({
         ":linux_x86_64": ["@shellcheck_linux_x86_64//:shellcheck"],
         ":macos_arm64": ["@shellcheck_darwin_arm64//:shellcheck"],
     }),
@@ -45,10 +45,12 @@ sh_test(
         ":linux_x86_64": {
             "SHELLCHECK": "$(location @shellcheck_linux_x86_64//:shellcheck)",
             "SETUP_SH": "$(location setup.sh)",
+            "SETUP_K8S_SH": "$(location setup-k8s.sh)",
         },
         ":macos_arm64": {
             "SHELLCHECK": "$(location @shellcheck_darwin_arm64//:shellcheck)",
             "SETUP_SH": "$(location setup.sh)",
+            "SETUP_K8S_SH": "$(location setup-k8s.sh)",
         },
     }),
 )
diff --git a/deployments/brev/README-k8s.md b/deployments/brev/README-k8s.md
new file mode 100644
index 000000000..851bbb9d8
--- /dev/null
+++ b/deployments/brev/README-k8s.md
@@ -0,0 +1,167 @@
+<!--
+SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+
+SPDX-License-Identifier: Apache-2.0
+-->
+
+# OSMO Brev Deployment (Kubernetes Mode)
+
+[![NVIDIA-OSMO](https://img.shields.io/badge/NVIDIA-OSMO-76b900?logo=nvidia)](https://github.com/NVIDIA/OSMO)
+[![Deploy on Brev](https://brev-assets.s3.us-west-1.amazonaws.com/nv-lb-dark.svg)](https://brev.nvidia.com/launchable/deploy?launchableID=env-3CPEGDFbjVhuXAffQYBEDd9qOi9)
+
+The OSMO Brev Kubernetes deployment provides a pre-configured OSMO instance running in the cloud, allowing you to quickly try OSMO without setting up local infrastructure. This deployment uses a [Brev.dev](https://brev.dev) cloud instance in **Single-node Kubernetes** mode with the [OSMO local deployment](https://nvidia.github.io/OSMO/main/deployment_guide/appendix/deploy_local.html) pre-installed.
+
+This is an alternative to the [VM mode deployment](README.md) — instead of installing Docker, KIND, and nvkind inside a VM, this mode uses Brev's pre-configured MicroK8s cluster with GPU support.
+
+> The Brev deployment is for evaluation purposes only and is not recommended for production use as it lacks authentication and has limited resources.
+
+## Compute requirements
+
+- NVIDIA Container Toolkit (>=1.18.1)
+- NVIDIA Driver Version (>=575)
+
+## Accessing the Brev Deployment
+
+### Web UI Access
+
+The OSMO Web UI is available through a secure Brev link exposed from your instance:
+
+1. Log in to your Brev console at https://console.brev.dev
+2. Navigate to your OSMO instance
+3. Select "Access"
+4. Under `Using Secure Links`, click `Share a Service` and enter port `30080`
+5. Click on the "Shareable URL" for port `30080`
+
+## [Optional] Local CLI Setup
+
+To use the OSMO CLI and UI from your local machine, you'll need to set up port forwarding and install the necessary tools.
+
+### Step 1: Install Brev CLI
+
+Follow instructions [here](https://docs.nvidia.com/brev/latest/brev-cli.html#installation-instructions). Be sure to `brev login`.
+
+### Step 2: Set Up Port Forwarding
+
+Forward ports from your Brev instance to your local machine. Port 30080 provides access to the OSMO API and Web UI. Port 30035 provides access to the LocalStack S3 storage backend, which is required for dataset download and upload via the CLI.
+
+You can find your instance's IP address at the top of the deployment page.
+
+You can see your username in the Brev Console:
+
+1. Log in to your Brev console at https://console.brev.dev
+2. Navigate to your OSMO instance
+3. Select "Logs"
+4. Look at the output of "Script Logs". You should see `Current user: [brev instance username]`
+
+```bash
+# Find your instance name with brev ls
+sudo ssh -i ~/.brev/brev.pem -p 22 -L 80:localhost:30080 <username>@[your instance IP]
+```
+
+If you see `Permission denied (publickey)` it may be because:
+
+- You did not log in using `brev login`
+
+### Step 3: Set Up Networking
+
+Add host entries so that the OSMO CLI and browser can reach the cluster services via localhost:
+
+```bash
+echo "127.0.0.1 quick-start.osmo" | sudo tee -a /etc/hosts
+echo "127.0.0.1 localstack-s3.osmo" | sudo tee -a /etc/hosts
+```
+
+`quick-start.osmo` allows you to visit the Web UI at `http://quick-start.osmo` in your browser. `localstack-s3.osmo` allows the OSMO CLI to reach the S3 storage backend for dataset download and upload.
+
+### Step 4: Install OSMO CLI
+
+Download and install the OSMO command-line interface:
+
+```bash
+curl -fsSL https://raw.githubusercontent.com/NVIDIA/OSMO/refs/heads/main/install.sh | bash
+```
+
+### Step 5: Log In to OSMO
+
+Authenticate with the OSMO instance through your port forward:
+
+```bash
+osmo login http://quick-start.osmo --method=dev --username=testuser
+```
+
+### Step 6: Set Dataset Credential
+
+Register the storage credential so the CLI can download and upload datasets. The setup script saves this command on the Brev node during installation. In a separate terminal, retrieve and run it:
+
+```bash
+ssh -i ~/.brev/brev.pem <username>@[your instance IP] 'cat ~/osmo-deployment/set-credential.sh' | bash
+```
+
+## Next Steps
+
+Visit the [User Guide](https://nvidia.github.io/OSMO/main/user_guide/getting_started/next_steps.html#getting-started-next-steps) for tutorials on submitting workflows, interactive development, distributed training, and more.
+
+## Additional Resources
+
+- [User Guide](https://nvidia.github.io/OSMO/main/user_guide/)
+- [Deployment Guide](https://nvidia.github.io/OSMO/main/deployment_guide/)
+- [OSMO GitHub Repository](https://github.com/nvidia/osmo)
+- [Brev Documentation](https://docs.brev.dev)
+
+## Cleanup
+
+Close the port-forward session with:
+
+```bash
+sudo kill -9 $(sudo lsof -ti:80)
+```
+
+Delete your Brev instance through the Brev console or CLI:
+
+```bash
+brev delete [your instance name]
+```
+
+# Deploying Custom OSMO Chart
+
+1. Build and push your quick-start chart to the registry.
+
+2. Go to [brev.nvidia.com](https://brev.nvidia.com) and create a new environment.
+   - Select **Single-node Kubernetes** mode.
+   - **L40S 1xGPU** on MassedCompute works well, but any L40 or L40S instance should work.
+
+3. Wait for the node to finish starting up.
+
+4. Shell into the instance:
+
+   ```bash
+   brev shell <your node name>
+   ```
+
+5. Download the setup script:
+
+   ```bash
+   curl -o setup-k8s.sh https://raw.githubusercontent.com/NVIDIA/OSMO/main/deployments/brev/setup-k8s.sh && chmod +x setup-k8s.sh
+   ```
+
+6. Edit `setup-k8s.sh` to install your version and use your registry key
+
+7. Run the setup script:
+
+   ```bash
+   ./setup-k8s.sh
+   ```
+
+Once you are complete, you can follow the instructions above to access your Brev instance.
diff --git a/deployments/brev/setup-k8s.sh b/deployments/brev/setup-k8s.sh
new file mode 100644
index 000000000..38002a4db
--- /dev/null
+++ b/deployments/brev/setup-k8s.sh
@@ -0,0 +1,351 @@
+#!/bin/bash
+
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+# SPDX-License-Identifier: Apache-2.0
+
+set -e
+
+# OSMO Single-node Kubernetes Deployment Script
+# For use with Brev's "Single-node Kubernetes" mode which provides:
+#   MicroK8s, GPU Operator, DNS, Hostpath Storage, kubectl, Helm 3
+# This script installs KAI Scheduler and OSMO on the pre-configured cluster.
+
+echo "=================================================="
+echo "OSMO Single-node K8s Deployment Script"
+echo "=================================================="
+echo ""
+
+print_status() {
+    echo "[INFO] $1"
+}
+
+print_warning() {
+    echo "[WARN] $1"
+}
+
+print_error() {
+    echo "[ERROR] $1"
+}
+
+command_exists() {
+    command -v "$1" >/dev/null 2>&1
+}
+
+# ============================================
+# Version Constants
+# ============================================
+KAI_SCHEDULER_VERSION="v0.13.4"
+
+# LocalStack S3 object storage settings
+LOCALSTACK_S3_HOST="localstack-s3.osmo"
+LOCALSTACK_S3_PORT="4566"
+LOCALSTACK_S3_OVERRIDE_URL="http://${LOCALSTACK_S3_HOST}:${LOCALSTACK_S3_PORT}"
+LOCALSTACK_S3_ENDPOINT="s3://osmo"
+LOCALSTACK_S3_ACCESS_KEY_ID="test"
+LOCALSTACK_S3_ACCESS_KEY="test"
+LOCALSTACK_S3_REGION="us-east-1"
+
+# ============================================
+# Step 0: Validate Prerequisites
+# ============================================
+print_status "Validating prerequisites..."
+
+# Verify kubectl works
+if ! kubectl get nodes >/dev/null 2>&1; then
+    print_error "kubectl cannot reach the cluster. Is MicroK8s running?"
+    exit 1
+fi
+print_status "Cluster is reachable"
+
+# Show cluster info
+kubectl get nodes -o wide
+echo ""
+
+# Check for GPU availability
+print_status "Checking GPU availability..."
+if kubectl get nodes -o json | python3 -c "
+import json, sys
+nodes = json.load(sys.stdin)['items']
+for n in nodes:
+    gpus = n.get('status', {}).get('capacity', {}).get('nvidia.com/gpu', '0')
+    print(f\"  Node {n['metadata']['name']}: {gpus} GPU(s)\")
+    if int(gpus) > 0:
+        sys.exit(0)
+sys.exit(1)
+" 2>/dev/null; then
+    print_status "GPU(s) detected in cluster"
+else
+    print_warning "No GPUs detected yet — GPU operator may still be initializing"
+fi
+
+# Check storage class
+print_status "Available storage classes:"
+kubectl get storageclass 2>/dev/null || print_warning "No storage classes found"
+
+# Detect the storage class to use
+STORAGE_CLASS=""
+if kubectl get storageclass microk8s-hostpath >/dev/null 2>&1; then
+    STORAGE_CLASS="microk8s-hostpath"
+elif kubectl get storageclass standard >/dev/null 2>&1; then
+    STORAGE_CLASS="standard"
+else
+    # Use whatever default storage class exists
+    STORAGE_CLASS=$(kubectl get storageclass -o jsonpath='{.items[?(@.metadata.annotations.storageclass\.kubernetes\.io/is-default-class=="true")].metadata.name}' 2>/dev/null || echo "")
+    if [ -z "$STORAGE_CLASS" ]; then
+        print_warning "No default storage class found — PVCs may fail"
+        STORAGE_CLASS="microk8s-hostpath"
+    fi
+fi
+print_status "Using storage class: $STORAGE_CLASS"
+
+# ============================================
+# Step 1: Label node for OSMO scheduling
+# ============================================
+# The OSMO Helm chart uses nodeSelector with node_group labels to place pods.
+# In a multi-node KIND cluster these map to separate worker nodes (service,
+# data, compute, ingress, kai-scheduler). On a single-node cluster we label
+# the only node as "compute" and point every chart nodeSelector at it.
+NODE_NAME=$(kubectl get nodes -o jsonpath='{.items[0].metadata.name}')
+print_status "Labeling node $NODE_NAME with node_group=compute..."
+kubectl label node "$NODE_NAME" node_group=compute --overwrite
+
+# ============================================
+# Step 2: Install KAI Scheduler
+# ============================================
+print_status "Installing KAI Scheduler..."
+
+helm upgrade --install kai-scheduler \
+  oci://ghcr.io/kai-scheduler/kai-scheduler/kai-scheduler \
+  --version ${KAI_SCHEDULER_VERSION} \
+  --create-namespace -n kai-scheduler \
+  --set "scheduler.additionalArgs[0]=--default-staleness-grace-period=-1s" \
+  --set "scheduler.additionalArgs[1]=--update-pod-eviction-condition=true"
+
+print_status "Waiting for KAI Scheduler pods..."
+kubectl -n kai-scheduler wait --for=condition=Available deployment --all --timeout=300s
+print_status "KAI Scheduler ready"
+
+# ============================================
+# Step 3: Install OSMO
+# ============================================
+print_status "Installing OSMO (this may take 5-10 minutes)..."
+
+helm repo add osmo https://helm.ngc.nvidia.com/nvidia/osmo
+helm repo update
+
+# All nodeSelectors are pointed at "compute" to match the label applied in
+# Step 1. The chart defaults use different node_group values (service, data,
+# ingress) for multi-node clusters — here we unify them onto the single node.
+helm upgrade --install osmo osmo/quick-start \
+  --namespace osmo \
+  --create-namespace \
+  --set global.nodeSelector.node_group=compute \
+  --set service.services.postgres.nodeSelector.node_group=compute \
+  --set service.services.postgres.storageClassName="${STORAGE_CLASS}" \
+  --set service.services.redis.nodeSelector.node_group=compute \
+  --set service.services.redis.storageClassName="${STORAGE_CLASS}" \
+  --set service.services.localstackS3.nodeSelector.node_group=compute \
+  --set ingress-nginx.controller.nodeSelector.node_group=compute \
+  --set ingress-nginx.controller.admissionWebhooks.nodeSelector.node_group=compute \
+  --set global.objectStorage.endpoint="${LOCALSTACK_S3_ENDPOINT}" \
+  --set global.objectStorage.overrideUrl="${LOCALSTACK_S3_OVERRIDE_URL}" \
+  --set global.objectStorage.accessKeyId="${LOCALSTACK_S3_ACCESS_KEY_ID}" \
+  --set global.objectStorage.accessKey="${LOCALSTACK_S3_ACCESS_KEY}" \
+  --set global.objectStorage.region="${LOCALSTACK_S3_REGION}" \
+  --set web-ui.services.ui.hostname="" \
+  --set service.services.service.hostname="" \
+  --set router.services.service.hostname="" \
+  --wait \
+  --timeout 10m
+
+print_status "OSMO installed successfully"
+
+# Verify all pods are running
+print_status "Verifying OSMO pods..."
+kubectl get pods --namespace osmo
+
+# ============================================
+# Step 4: Restart KAI Scheduler
+# ============================================
+# MicroK8s service account tokens can be stale at initial pod creation, causing
+# "Unauthorized" errors when KAI tries to list PodGroups. A rollout restart
+# forces fresh tokens and resolves the issue.
+print_status "Restarting KAI Scheduler to refresh service account tokens..."
+kubectl rollout restart deployment/kai-scheduler-default -n kai-scheduler
+kubectl -n kai-scheduler rollout status deployment/kai-scheduler-default --timeout=60s
+print_status "KAI Scheduler restarted"
+
+OSMO_GATEWAY_PORT=$(kubectl get svc -n osmo quick-start -o jsonpath='{.spec.ports[0].nodePort}' 2>/dev/null || echo "30080")
+OSMO_API="http://localhost:${OSMO_GATEWAY_PORT}"
+
+# Wait for the API to be reachable
+print_status "Waiting for OSMO API at ${OSMO_API}..."
+until curl -sf "${OSMO_API}/api/configs/pod_template/default_compute" -H "x-osmo-user: testuser" >/dev/null 2>&1; do
+    echo "  API not ready yet, waiting..."
+    sleep 5
+done
+print_status "OSMO API is reachable"
+
+# ============================================
+# Step 5: Add shared_memory pod template
+# ============================================
+print_status "Adding shared_memory pod template for /dev/shm..."
+
+curl -sf -X PUT "${OSMO_API}/api/configs/pod_template/shared_memory" \
+  -H "Content-Type: application/json" \
+  -H "x-osmo-user: testuser" \
+  -d '{
+    "configs": {
+      "spec": {
+        "volumes": [
+          {
+            "name": "shm",
+            "emptyDir": {
+              "medium": "Memory",
+              "sizeLimit": "64Gi"
+            }
+          }
+        ],
+        "containers": [
+          {
+            "name": "{{USER_CONTAINER_NAME}}",
+            "volumeMounts": [
+              {
+                "name": "shm",
+                "mountPath": "/dev/shm"
+              }
+            ]
+          }
+        ]
+      }
+    },
+    "description": "Add shared_memory pod template for /dev/shm mounting"
+  }'
+
+print_status "shared_memory pod template created"
+
+# Add shared_memory to the default pool's common_pod_template list
+print_status "Adding shared_memory to default pool..."
+
+CURRENT_COMMON_POD_TEMPLATE=$(
+  curl -sf "${OSMO_API}/api/configs/pool/default" \
+    -H "x-osmo-user: testuser" \
+  | python3 -c '
+import json, sys
+data = json.load(sys.stdin)
+templates = data.get("common_pod_template") or []
+if "shared_memory" not in templates:
+    templates.append("shared_memory")
+print(json.dumps(templates))
+'
+)
+
+curl -sf -X PATCH "${OSMO_API}/api/configs/pool/default" \
+  -H "Content-Type: application/json" \
+  -H "x-osmo-user: testuser" \
+  -d "{
+    \"configs_dict\": {
+      \"common_pod_template\": ${CURRENT_COMMON_POD_TEMPLATE}
+    },
+    \"description\": \"Add shared_memory pod template to default pool\"
+  }"
+
+print_status "Default pool updated with shared_memory pod template"
+
+# ============================================
+# Step 6: /etc/hosts
+# ============================================
+if ! grep -q "${LOCALSTACK_S3_HOST}" /etc/hosts; then
+    print_status "Adding ${LOCALSTACK_S3_HOST} to /etc/hosts..."
+    echo "127.0.0.1 ${LOCALSTACK_S3_HOST}" | sudo tee -a /etc/hosts
+else
+    print_status "${LOCALSTACK_S3_HOST} already in /etc/hosts"
+fi
+
+# ============================================
+# Step 7: Install OSMO CLI
+# ============================================
+print_status "Installing OSMO CLI..."
+
+curl -fsSL https://raw.githubusercontent.com/NVIDIA/OSMO/refs/heads/main/install.sh -o /tmp/install-osmo.sh
+chmod +x /tmp/install-osmo.sh
+sudo bash /tmp/install-osmo.sh
+
+# Add OSMO to PATH if not already there
+if [[ ":$PATH:" != *":$HOME/.osmo/bin:"* ]]; then
+    export PATH="$HOME/.osmo/bin:$PATH"
+    # shellcheck disable=SC2016
+    echo 'export PATH="$HOME/.osmo/bin:$PATH"' >> ~/.bashrc
+fi
+
+# ============================================
+# Step 8: Log In and Set Credential
+# ============================================
+print_status "Logging in to OSMO..."
+
+osmo login "${OSMO_API}" --method=dev --username=testuser
+
+print_status "Setting data credential for LocalStack S3..."
+
+osmo credential set osmo --type DATA --payload \
+  access_key_id="${LOCALSTACK_S3_ACCESS_KEY_ID}" \
+  access_key="${LOCALSTACK_S3_ACCESS_KEY}" \
+  endpoint="${LOCALSTACK_S3_ENDPOINT}" \
+  override_url="${LOCALSTACK_S3_OVERRIDE_URL}" \
+  region="${LOCALSTACK_S3_REGION}"
+
+# Save the credential command for remote workstation setup
+mkdir -p ~/osmo-deployment
+cat > ~/osmo-deployment/set-credential.sh <<CRED_EOF
+#!/bin/bash
+osmo credential set osmo --type DATA --payload \\
+  access_key_id="${LOCALSTACK_S3_ACCESS_KEY_ID}" \\
+  access_key="${LOCALSTACK_S3_ACCESS_KEY}" \\
+  endpoint="${LOCALSTACK_S3_ENDPOINT}" \\
+  override_url="${LOCALSTACK_S3_OVERRIDE_URL}" \\
+  region="${LOCALSTACK_S3_REGION}"
+CRED_EOF
+chmod +x ~/osmo-deployment/set-credential.sh
+
+print_status "Data credential set successfully"
+
+# ============================================
+# Success Message
+# ============================================
+echo ""
+echo "=================================================="
+echo "OSMO Deployment Complete! (Single-node K8s)"
+echo "=================================================="
+echo ""
+print_status "System Information:"
+print_status "  • Storage Class: $STORAGE_CLASS"
+print_status "  • OSMO API: $OSMO_API"
+print_status "  • LocalStack S3: ${LOCALSTACK_S3_OVERRIDE_URL}"
+echo ""
+print_status "GPU status:"
+kubectl get nodes -o json | python3 -c "
+import json, sys
+nodes = json.load(sys.stdin)['items']
+for n in nodes:
+    gpus = n.get('status', {}).get('capacity', {}).get('nvidia.com/gpu', '0')
+    print(f'  Node {n[\"metadata\"][\"name\"]}: {gpus} GPU(s)')
+" 2>/dev/null || print_warning "Could not check GPU status"
+echo ""
+print_status "Next Steps:"
+print_status "  1. See deployment README: https://github.com/nvidia/osmo/tree/main/deployments/brev/README-k8s.md"
+print_status "  2. Follow getting started guide: https://nvidia.github.io/OSMO/main/user_guide/getting_started/next_steps.html"
+echo ""
+echo "=================================================="
diff --git a/deployments/brev/shellcheck_test.sh b/deployments/brev/shellcheck_test.sh
index 66ce4107d..34f1b9ee6 100755
--- a/deployments/brev/shellcheck_test.sh
+++ b/deployments/brev/shellcheck_test.sh
@@ -18,3 +18,4 @@
 set -euo pipefail
 
 "$SHELLCHECK" --severity=warning "$SETUP_SH"
+"$SHELLCHECK" --severity=warning "$SETUP_K8S_SH"