Helion support and Lambda Cloud Deploy Actions

.github/workflows/ci.yml

@@ -46,3 +46,14 @@ jobs:
 
       - name: ort identity (cpu)
         run: cargo test -p kernelport-backend-ort --test identity
+
+  python-scripts:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: install PyYAML
+        run: pip install pyyaml
+
+      - name: manifest_to_serve_args tests
+        run: python tests/test_manifest_serve_args.py

.github/workflows/deploy-lambda.yml

@@ -0,0 +1,127 @@
+# Deploy KernelPort (LuxTTS) to Lambda Cloud: build/push images to GHCR, launch GPU instance.
+# Trigger: workflow_dispatch. Requires repository secrets: LAMBDA_CLOUD_API_KEY, HUGGINGFACE_HUB_TOKEN.
+# You must add an SSH key in Lambda Cloud (https://cloud.lambda.ai/ssh-keys) and pass its name as input.
+name: Deploy to Lambda Cloud
+
+on:
+  workflow_dispatch:
+    inputs:
+      instance_type_name:
+        description: "Lambda instance type (e.g. gpu_1x_a100)"
+        required: true
+        default: "gpu_1x_a100"
+      region_name:
+        description: "Lambda region (e.g. us-tx-1)"
+        required: true
+        default: "us-tx-1"
+      ssh_key_name:
+        description: "Name of SSH key already added in Lambda Cloud"
+        required: true
+
+env:
+  LAMBDA_API_BASE: "https://cloud.lambdalabs.com/api/v1"
+
+jobs:
+  build-push:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      - name: Log in to GHCR
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build and push kernelport (GPU)
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          file: ./Dockerfile.gpu
+          push: true
+          tags: ghcr.io/${{ github.repository_owner }}/kernelport:gpu,ghcr.io/${{ github.repository_owner }}/kernelport:gpu-${{ github.sha }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+
+      - name: Build and push kernelport-luxtts (GPU)
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          file: ./Dockerfile.luxtts
+          push: true
+          tags: ghcr.io/${{ github.repository_owner }}/kernelport-luxtts:gpu,ghcr.io/${{ github.repository_owner }}/kernelport-luxtts:gpu-${{ github.sha }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+
+  launch:
+    needs: build-push
+    runs-on: ubuntu-latest
+    steps:
+      - name: List instance types (optional)
+        run: |
+          curl -sSf -u "${{ secrets.LAMBDA_CLOUD_API_KEY }}:" \
+            "${{ env.LAMBDA_API_BASE }}/instance-types" | jq -r '.data | keys[]' | head -20
+
+      - name: Launch instance
+        id: launch
+        run: |
+          body=$(jq -n \
+            --arg region "${{ github.event.inputs.region_name }}" \
+            --arg type "${{ github.event.inputs.instance_type_name }}" \
+            --arg key "${{ github.event.inputs.ssh_key_name }}" \
+            '{region_name: $region, instance_type_name: $type, ssh_key_names: [$key]}')
+          resp=$(curl -sSf -u "${{ secrets.LAMBDA_CLOUD_API_KEY }}:" \
+            -X POST "${{ env.LAMBDA_API_BASE }}/instance-operations/launch" \
+            -H "Content-Type: application/json" \
+            -d "$body")
+          echo "instance_ids=$(echo "$resp" | jq -r '.data.instance_ids[0]')" >> $GITHUB_OUTPUT
+          echo "$resp" | jq .
+
+      - name: Poll until instance is active
+        id: poll
+        run: |
+          id="${{ steps.launch.outputs.instance_ids }}"
+          for i in $(seq 1 30); do
+            resp=$(curl -sSf -u "${{ secrets.LAMBDA_CLOUD_API_KEY }}:" \
+              "${{ env.LAMBDA_API_BASE }}/instances/$id")
+            status=$(echo "$resp" | jq -r '.data.status')
+            ip=$(echo "$resp" | jq -r '.data.ip // empty')
+            echo "Attempt $i: status=$status ip=$ip"
+            if [ "$status" = "active" ] && [ -n "$ip" ]; then
+              echo "instance_ip=$ip" >> $GITHUB_OUTPUT
+              echo "Instance is up at $ip"
+              exit 0
+            fi
+            sleep 20
+          done
+          echo "Instance did not become active in time"
+          exit 1
+
+      - name: Smoke test (grpcurl)
+        continue-on-error: true
+        run: |
+          ip="${{ steps.poll.outputs.instance_ip }}"
+          if [ -z "$ip" ]; then exit 0; fi
+          # Instance may not have the stack running yet; user can SSH and run docker compose
+          if command -v grpcurl >/dev/null 2>&1; then
+            grpcurl -plaintext -connect-timeout 5 "$ip:50051" list || true
+          else
+            echo "grpcurl not installed; skip smoke test. Endpoint: $ip:50051"
+          fi
+
+      - name: Summary
+        run: |
+          echo "## Lambda Cloud instance" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "- **Instance ID**: ${{ steps.launch.outputs.instance_ids }}" >> $GITHUB_STEP_SUMMARY
+          echo "- **IP**: ${{ steps.poll.outputs.instance_ip }}" >> $GITHUB_STEP_SUMMARY
+          echo "- **gRPC endpoint**: ${{ steps.poll.outputs.instance_ip }}:50051 (after you deploy the stack on the instance)" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "SSH into the instance and run your stack (e.g. docker compose). Set \`HUGGINGFACE_HUB_TOKEN\` when running the LuxTTS worker." >> $GITHUB_STEP_SUMMARY

.gitignore

@@ -17,6 +17,9 @@ target
 *.onnx
 !models/identity.onnx
 
+# Editor / IDE
+.vscode/
+
 # RustRover
 #  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore

.pre-commit-config.yaml

@@ -7,3 +7,15 @@ repos:
         language: system
         types: [rust]
         pass_filenames: false
+
+      - id: cargo-test
+        name: cargo test
+        entry: cargo test --all
+        language: system
+        pass_filenames: false
+
+      - id: python-scripts
+        name: Python script tests (manifest_to_serve_args)
+        entry: python tests/test_manifest_serve_args.py
+        language: system
+        pass_filenames: false

Cargo.toml

@@ -7,7 +7,7 @@ members = [
   "crates/kernelport-core",
   "crates/kernelport-runtime",
   "crates/kernelport-backend-ort",
+  "crates/kernelport-backend-helion",
   "crates/kernelport-proto",
   "crates/kernelport-server",
 ]
-

Dockerfile.helion

@@ -0,0 +1,32 @@
+FROM nvidia/cuda:12.2.0-runtime-ubuntu22.04
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        ca-certificates \
+        curl \
+        python3 \
+        python3-venv \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+
+ENV VIRTUAL_ENV=/opt/venv
+ENV PATH="/opt/venv/bin:/root/.cargo/bin:${PATH}"
+
+RUN uv venv /opt/venv \
+    && uv pip install \
+        "torch==2.9.*" \
+        --index-url https://download.pytorch.org/whl/cu126 \
+    && uv pip install \
+        helion \
+        grpcio \
+        grpcio-tools \
+        numpy
+
+WORKDIR /app
+COPY scripts/helion /app/scripts/helion
+COPY crates/kernelport-proto/src /app/crates/kernelport-proto/src
+
+EXPOSE 50061
+
+CMD ["python", "/app/scripts/helion/helion_worker.py", "--addr", "0.0.0.0:50061", "--device", "cuda"]

Dockerfile.helion.mock

@@ -0,0 +1,25 @@
+FROM python:3.11-slim
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        ca-certificates \
+        curl \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+
+ENV PATH="/root/.local/bin:/root/.cargo/bin:${PATH}"
+
+WORKDIR /app
+COPY scripts/helion /app/scripts/helion
+COPY crates/kernelport-proto/src /app/crates/kernelport-proto/src
+
+RUN /root/.local/bin/uv venv /app/.venv \
+    && . /app/.venv/bin/activate \
+    && /root/.local/bin/uv pip install numpy grpcio grpcio-tools
+
+ENV PATH="/app/.venv/bin:${PATH}"
+
+EXPOSE 50061
+
+CMD ["python", "/app/scripts/helion/mock/helion_worker_mock.py", "--addr", "0.0.0.0:50061"]

Dockerfile.luxtts

@@ -0,0 +1,38 @@
+# LuxTTS gRPC worker (Helion-style sidecar). Loads YatharthS/LuxTTS from HF.
+# Uses uv for consistency with Dockerfile.helion.
+FROM nvidia/cuda:12.2.0-runtime-ubuntu22.04
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        ca-certificates \
+        curl \
+        git \
+        python3 \
+        python3-venv \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+
+ENV VIRTUAL_ENV=/opt/venv
+ENV PATH="/opt/venv/bin:/root/.local/bin:${PATH}"
+
+WORKDIR /app
+
+# Clone LuxTTS so zipvoice.luxvoice is available; install its deps with uv
+RUN git clone --depth 1 https://github.com/ysharma3501/LuxTTS.git /app/LuxTTS
+RUN uv venv /opt/venv \
+    && uv pip install -r /app/LuxTTS/requirements.txt
+
+# gRPC and proto for kernelport InferenceService
+RUN uv pip install grpcio grpcio-tools
+
+COPY scripts/luxtts /app/scripts/luxtts
+COPY crates/kernelport-proto/src /app/crates/kernelport-proto/src
+
+ENV LUXTTS_REPO=/app/LuxTTS
+ENV KERNELPORT_REPO_ROOT=/app
+ENV PYTHONPATH=/app/LuxTTS:/app
+
+EXPOSE 50061
+
+CMD ["python", "/app/scripts/luxtts/luxtts_worker.py", "--addr", "0.0.0.0:50061", "--device", "cuda"]

Makefile

@@ -1,5 +1,5 @@
 .DEFAULT_GOAL := help
-.PHONY: fmt clippy test build check run help
+.PHONY: fmt clippy test test-python build check run help
 
 PROJECT_NAME := kernelport
 PROJECT_VERSION := $(shell cargo pkgid -p kernelport-server 2>/dev/null | sed -E 's/.*@//')
@@ -16,10 +16,13 @@ clippy:
 test:
 	cargo test --all
 
+test-python:
+	python3 tests/test_manifest_serve_args.py
+
 build:
 	cargo build --all
 
-check: fmt clippy test build
+check: fmt clippy test test-python build
 
 run:
 	RUST_LOG=info cargo run -p kernelport-server
@@ -28,9 +31,10 @@ help:
 	@printf "%s\n" \
 		"$(PROJECT_NAME) $(PROJECT_VERSION)" \
 		"Targets:" \
-		"  fmt    - Run rustfmt" \
-		"  clippy - Run clippy with warnings denied" \
-		"  test   - Run tests" \
-		"  build  - Build all crates" \
-		"  check  - Run fmt, clippy, test, build" \
-		"  run    - Run kernelport-server"
+		"  fmt         - Run rustfmt" \
+		"  clippy      - Run clippy with warnings denied" \
+		"  test        - Run Rust tests" \
+		"  test-python - Run Python script tests (manifest_to_serve_args)" \
+		"  build       - Build all crates" \
+		"  check       - Run fmt, clippy, test, test-python, build" \
+		"  run         - Run kernelport-server"