diff --git a/.github/actions/pr-permission-gate/action.yml b/.github/actions/pr-permission-gate/action.yml
index d6078207e1ef..4288a11bb9a3 100644
--- a/.github/actions/pr-permission-gate/action.yml
+++ b/.github/actions/pr-permission-gate/action.yml
@@ -4,8 +4,9 @@ inputs:
   required-level:
     description: Minimum permission level required (read|triage|write|maintain|admin)
     default: write
+
 runs:
-  using: "composite"
+  using: composite
   steps:
     - name: Check PR sender permission
       uses: actions/github-script@v7
diff --git a/.github/workflows/pr-sglang-sagemaker.yml b/.github/workflows/pr-sglang-sagemaker.yml
new file mode 100644
index 000000000000..9e8e03c17528
--- /dev/null
+++ b/.github/workflows/pr-sglang-sagemaker.yml
@@ -0,0 +1,229 @@
+name: PR - SGLang
+
+on:
+  pull_request:
+    branches:
+      - main
+    paths:
+      - "docker/sglang/**"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: pr-sglang-${{ github.event.pull_request.number }}
+  cancel-in-progress: true
+
+env:
+  SGLANG_VERSION: "0.5.5"
+  PYTHON_VERSION: "py312"
+  CUDA_VERSION: "cu129"
+  OS_VERSION: "ubuntu22.04"
+  FORCE_COLOR: "1"
+
+jobs:
+  check-changes:
+    runs-on: ubuntu-latest
+    outputs:
+      sglang-sagemaker: ${{ steps.changes.outputs.sglang-sagemaker }}
+    steps:
+      - uses: actions/checkout@v5
+      - uses: actions/setup-python@v6
+        with:
+          python-version: "3.12"
+      - uses: pre-commit/action@v3.0.1
+        with:
+          extra_args: --all-files
+      - name: Detect file changes
+        id: changes
+        uses: dorny/paths-filter@v3
+        with:
+          filters: |
+            sglang-sagemaker:
+              - "docker/sglang/Dockerfile"
+
+  build-sglang-image:
+    needs: [check-changes]
+    if: needs.check-changes.outputs.sglang-sagemaker == 'true'
+    runs-on:
+      - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }}
+        fleet:x86-build-runner
+    outputs:
+      image-uri: ${{ steps.image-uri-build.outputs.IMAGE_URI }}
+    steps:
+      - uses: actions/checkout@v5
+      - run: .github/scripts/runner_setup.sh
+      - run: .github/scripts/buildkitd.sh
+
+      - name: ECR login
+        uses: ./.github/actions/ecr-authenticate
+        with:
+          aws_region: ${{ vars.AWS_REGION }}
+          aws_account_id: ${{ vars.AWS_ACCOUNT_ID }}
+
+      - name: Resolve image URI for build
+        id: image-uri-build
+        run: |
+          IMAGE_URI=${{ vars.AWS_ACCOUNT_ID }}.dkr.ecr.${{ vars.AWS_REGION }}.amazonaws.com/ci:sglang-${{ env.SGLANG_VERSION }}-gpu-${{ env.PYTHON_VERSION }}-${{ env.CUDA_VERSION }}-${{ env.OS_VERSION }}-sagemaker-pr-${{ github.event.pull_request.number }}
+          echo "Image URI to build: $IMAGE_URI"
+          echo "IMAGE_URI=$IMAGE_URI" >> $GITHUB_ENV
+          echo "IMAGE_URI=$IMAGE_URI" >> $GITHUB_OUTPUT
+
+      - name: Build image
+        run: |
+          # base image: https://hub.docker.com/r/lmsysorg/sglang/tags
+          docker buildx build --progress plain \
+            --build-arg CACHE_REFRESH="$(date +"%Y-%m-%d")" \
+            --build-arg BASE_IMAGE="lmsysorg/sglang:v${{ env.SGLANG_VERSION }}-${{ env.CUDA_VERSION }}-amd64" \
+            --cache-to=type=inline \
+            --cache-from=type=registry,ref=${IMAGE_URI} \
+            --tag ${IMAGE_URI} \
+            --target sglang-sagemaker \
+            -f docker/sglang/Dockerfile .
+
+      - name: Container push
+        run: |
+          docker push ${IMAGE_URI}
+          docker rmi ${IMAGE_URI}
+
+  sglang-local-benchmark-test:
+    needs: [build-sglang-image]
+    if: needs.build-sglang-image.result == 'success'
+    runs-on:
+      - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }}
+        fleet:x86-g6xl-runner
+    steps:
+      - name: Checkout DLC source
+        uses: actions/checkout@v5
+
+      - name: Container pull
+        uses: ./.github/actions/ecr-authenticate
+        with:
+          aws_region: ${{ vars.AWS_REGION }}
+          aws_account_id: ${{ vars.AWS_ACCOUNT_ID }}
+          image_uri: ${{ needs.build-sglang-image.outputs.image-uri }}
+
+      - name: Setup for SGLang datasets
+        run: |
+          mkdir -p /tmp/sglang/dataset
+          if [ ! -f /tmp/sglang/dataset/ShareGPT_V3_unfiltered_cleaned_split.json ]; then
+              echo "Downloading ShareGPT dataset..."
+              wget -P /tmp/sglang/dataset https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
+          else
+              echo "ShareGPT dataset already exists. Skipping download."
+          fi
+
+      - name: Start container
+        run: |
+          CONTAINER_ID=$(docker run -d -it --rm --gpus=all \
+            -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \
+            -v /tmp/sglang/dataset:/dataset \
+            -p 30000:30000 \
+            -e SM_SGLANG_MODEL_PATH=Qwen/Qwen3-0.6B \
+            -e SM_SGLANG_REASONING_PARSER=qwen3 \
+            -e SM_SGLANG_HOST=127.0.0.1 \
+            -e SM_SGLANG_PORT=30000 \
+            -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \
+            ${{ needs.build-sglang-image.outputs.image-uri }})
+          echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV
+          echo "Waiting for container startup ..."
+          sleep 60s
+          docker logs ${CONTAINER_ID}
+
+      - name: Run SGLang tests
+        run: |
+          docker exec ${CONTAINER_ID} python3 -m sglang.bench_serving \
+          --backend sglang \
+          --host 127.0.0.1 --port 30000 \
+          --num-prompts 1000 \
+          --model Qwen/Qwen3-0.6B \
+          --dataset-name sharegpt \
+          --dataset-path /dataset/ShareGPT_V3_unfiltered_cleaned_split.json
+
+      - name: Cleanup container and images
+        if: always()
+        uses: ./.github/actions/container-cleanup
+        with:
+          container_id: ${{ env.CONTAINER_ID }}
+
+  sglang-lang-test:
+    needs: [build-sglang-image]
+    if: needs.build-sglang-image.result == 'success'
+    runs-on:
+      - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }}
+        fleet:x86-g6exl-runner
+    steps:
+      - name: Checkout DLC source
+        uses: actions/checkout@v5
+
+      - name: Container pull
+        uses: ./.github/actions/ecr-authenticate
+        with:
+          aws_region: ${{ vars.AWS_REGION }}
+          aws_account_id: ${{ vars.AWS_ACCOUNT_ID }}
+          image_uri: ${{ needs.build-sglang-image.outputs.image-uri }}
+
+      - name: Checkout SGLang tests
+        uses: actions/checkout@v5
+        with:
+          repository: sgl-project/sglang
+          ref: v${{ env.SGLANG_VERSION }}
+          path: sglang_source
+
+      - name: Start container
+        run: |
+          CONTAINER_ID=$(docker run -d -it --rm --gpus=all --entrypoint /bin/bash \
+            -v ${HOME}/.cache/huggingface:/root/.cache/huggingface \
+            -v ./sglang_source:/workdir --workdir /workdir \
+            -e HF_TOKEN=${{ secrets.HUGGING_FACE_HUB_TOKEN }} \
+            ${{ needs.build-sglang-image.outputs.image-uri }})
+          echo "CONTAINER_ID=$CONTAINER_ID" >> $GITHUB_ENV
+
+      - name: Setup for SGLang tests
+        run: |
+          docker exec ${CONTAINER_ID} sh -c '
+            set -eux
+
+            bash scripts/ci/ci_install_dependency.sh
+          '
+
+      - name: Run SGLang tests
+        run: |
+          docker exec ${CONTAINER_ID} sh -c '
+            set -eux
+            nvidia-smi
+
+            # Frontend Test
+            cd /workdir/test/lang
+            python3 run_suite.py --suite per-commit
+          '
+
+      - name: Cleanup container and images
+        if: always()
+        uses: ./.github/actions/container-cleanup
+        with:
+          container_id: ${{ env.CONTAINER_ID }}
+
+  sglang-sagemaker-test:
+    needs: [build-sglang-image]
+    if: needs.build-sglang-image.result == 'success'
+    runs-on:
+      - codebuild-runner-${{ github.run_id }}-${{ github.run_attempt }}
+        fleet:default-runner
+    steps:
+      - name: Checkout DLC source
+        uses: actions/checkout@v5
+
+      - name: Install test dependencies
+        run: |
+          uv venv
+          source .venv/bin/activate
+
+          uv pip install -r test/requirements.txt
+          uv pip install -r test/sglang/sagemaker/requirements.txt
+
+      - name: Run sagemaker tests
+        run: |
+          source .venv/bin/activate
+          cd test/
+          python3 -m pytest -vs -rA --image-uri ${{ needs.build-sglang-image.outputs.image-uri }} sglang/sagemaker
diff --git a/.gitignore b/.gitignore
index 126c4416f381..2a1f55c8f33e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -3,4 +3,5 @@ __pycache__
 .idea
 *.pyc
 .venv
-.ruff_cache
\ No newline at end of file
+.ruff_cache
+.pytest_cache
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index ec83d90cb8bf..4d721708888c 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -1,6 +1,6 @@
 default_install_hook_types:
-  - pre-commit
   - commit-msg
+  - pre-commit
 
 default_stages:
   - pre-commit
@@ -16,44 +16,44 @@ repos:
     rev: v6.0.0
     hooks:
   # ------------------------------- 🌳 Git Tools ------------------------------- #
-    - id: check-merge-conflict
-      name: "🌳 git · Detect conflict markers"
-    - id: forbid-new-submodules
-      name: "🌳 git · Prevent submodule creation"
-    - id: no-commit-to-branch
-      name: "🌳 git · Protect main branches"
-      args: ["--branch", "main", "--branch", "master"]
+      - id: check-merge-conflict
+        name: "🌳 git · Detect conflict markers"
+      - id: forbid-new-submodules
+        name: "🌳 git · Prevent submodule creation"
+      - id: no-commit-to-branch
+        name: "🌳 git · Protect main branches"
+        args: ["--branch", "main", "--branch", "master"]
   # ---------------------------- 📁 Filesystem Tools --------------------------- #
-    - id: check-executables-have-shebangs
-      name: "📁 filesystem/⚙️ exec · Verify shebang presence"
-    - id: check-shebang-scripts-are-executable
-      name: "📁 filesystem/⚙️ exec · Verify script permissions"
-    - id: check-symlinks
-      name: "📁 filesystem/🔗 symlink · Check symlink validity"
-    - id: destroyed-symlinks
-      name: "📁 filesystem/🔗 symlink · Detect broken symlinks"
-    - id: requirements-txt-fixer
-      name: "📁 filesystem/📝 file · Validate requirements.txt"
-    - id: trailing-whitespace
-      name: "📁 filesystem/📝 file · Validate empty whitespaces"
+      - id: check-executables-have-shebangs
+        name: "📁 filesystem/⚙️ exec · Verify shebang presence"
+      - id: check-shebang-scripts-are-executable
+        name: "📁 filesystem/⚙️ exec · Verify script permissions"
+      - id: check-symlinks
+        name: "📁 filesystem/🔗 symlink · Check symlink validity"
+      - id: destroyed-symlinks
+        name: "📁 filesystem/🔗 symlink · Detect broken symlinks"
+      - id: requirements-txt-fixer
+        name: "📁 filesystem/📝 file · Validate requirements.txt"
+      - id: trailing-whitespace
+        name: "📁 filesystem/📝 file · Validate empty whitespaces"
   # ----------------------------- 🔒 Security Tools ---------------------------- #
-    - id: check-json
-      name: "📁 filesystem/🔧 config · Validate JSON file"
-    - id: check-yaml
-      name: "📁 filesystem/🔧 config · Validate YAML file"
-    - id: detect-aws-credentials
-      name: "🔒 security · Detect aws credentials"
-      args:
-        - --allow-missing-credentials
-    - id: detect-private-key
-      name: "🔒 security · Detect private keys"
+      - id: check-json
+        name: "📁 filesystem/🔧 config · Validate JSON file"
+      - id: check-yaml
+        name: "📁 filesystem/🔧 config · Validate YAML file"
+      - id: detect-aws-credentials
+        name: "🔒 security · Detect aws credentials"
+        args:
+          - --allow-missing-credentials
+      - id: detect-private-key
+        name: "🔒 security · Detect private keys"
 
 
   - repo: https://github.com/gitleaks/gitleaks
     rev: v8.29.0
     hooks:
-    - id: gitleaks
-      name: "🔒 security · Detect hardcoded secrets"
+      - id: gitleaks
+        name: "🔒 security · Detect hardcoded secrets"
 
   # --------------------------- 🔍 Code Quality Tools -------------------------- #
 
@@ -61,80 +61,78 @@ repos:
   - repo: https://github.com/crate-ci/typos
     rev: v1.38.1
     hooks:
-    - id: typos
-      name: "✏️ typos · Checking typos"
-      args: [--force-exclude]
+      - id: typos
+        name: "✏️ typos · Checking typos"
+        args: [--force-exclude]
 
 
   ### Python Tools ###
   - repo: https://github.com/astral-sh/ruff-pre-commit
     rev: v0.14.3
     hooks:
-    - id: ruff-format
-      name: "🐍 python · Format python"
-    - id: ruff-check
-      name: "🐍 python · Lint python"
+      - id: ruff-format
+        name: "🐍 python · Format python"
+      - id: ruff-check
+        name: "🐍 python · Lint python"
+        args: ["--fix"]
+      - id: ruff-check
+        name: "🐍 python · Sort python imports"
+        args: ["check", "--select", "I", "--fix"]
 
 
   ### Workflow Validation ###
   - repo: https://github.com/python-jsonschema/check-jsonschema
     rev: 0.35.0
     hooks:
-    - id: check-github-workflows
-      name: "🐙 github-actions · Validate gh workflow files"
-      args: ["--verbose"]
-
-  - repo: https://github.com/pycqa/isort
-    rev: 7.0.0
-    hooks:
-      - id: isort
-        name: "🐍 python · Sort python packages"
+      - id: check-github-workflows
+        name: "🐙 github-actions · Validate gh workflow files"
+        args: ["--verbose"]
 
   - repo: https://github.com/rhysd/actionlint
     rev: v1.7.8
     hooks:
-    - id: actionlint
-      name: "🐙 github-actions · Lint gh workflow files"
-      args: ["-shellcheck=enable=all"]
+      - id: actionlint
+        name: "🐙 github-actions · Lint gh workflow files"
+        args: ["-shellcheck=enable=all"]
 
 
   ### Markdown ###
   - repo: https://github.com/hukkin/mdformat
     rev: 1.0.0
     hooks:
-    - id: mdformat
-      name: "📝 markdown · Format markdown"
-      additional_dependencies:
-      - mdformat-gfm
-      - mdformat-ruff
-      - ruff
+      - id: mdformat
+        name: "📝 markdown · Format markdown"
+        additional_dependencies:
+        - mdformat-gfm
+        - mdformat-ruff
+        - ruff
 
   - repo: https://github.com/igorshubovych/markdownlint-cli
     rev: v0.45.0
     hooks:
-    - id: markdownlint
-      name: "📝 markdown · Lint markdown"
-      args: [--fix]
+      - id: markdownlint
+        name: "📝 markdown · Lint markdown"
+        args: [--fix]
 
 
   ### Docker Tools ###
   - repo: https://github.com/reteps/dockerfmt
     rev: v0.3.9
     hooks:
-    - id: dockerfmt
-      name: "🐳 docker · Format docker"
-      args:
-        - --indent=2
-        - --write
+      - id: dockerfmt
+        name: "🐳 docker · Format docker"
+        args:
+          - --indent=2
+          - --write
 
 
   ### Shell ###
   - repo: https://github.com/scop/pre-commit-shfmt
     rev: v3.12.0-2 # Use the latest stable revision
     hooks:
-    - id: shfmt
-      name: "🐚 shell · Format shell scripts"
-      args: ["-s"]
+      - id: shfmt
+        name: "🐚 shell · Format shell scripts"
+        args: ["-s"]
 
 
   # --------------------------- ✍️ Git Commit Quality -------------------------- #
@@ -142,15 +140,15 @@ repos:
   ### Commit Message Standards ###
   - repo: local
     hooks:
-    - id: signoff-commit
-      name: Sign-off Commit
-      entry: bash
-      args:
-        - -c
-        - |
-          if ! grep -q "^Signed-off-by: $(git config user.name) <$(git config user.email)>" "$(git rev-parse --git-path COMMIT_EDITMSG)"; then
-            printf "\nSigned-off-by: $(git config user.name) <$(git config user.email)>\n" >> "$(git rev-parse --git-path COMMIT_EDITMSG)"
-          fi
-      language: system
-      verbose: true
-      stages: [commit-msg]
+      - id: signoff-commit
+        name: Sign-off Commit
+        entry: bash
+        args:
+          - -c
+          - |
+            if ! grep -q "^Signed-off-by: $(git config user.name) <$(git config user.email)>" "$(git rev-parse --git-path COMMIT_EDITMSG)"; then
+              printf "\nSigned-off-by: $(git config user.name) <$(git config user.email)>\n" >> "$(git rev-parse --git-path COMMIT_EDITMSG)"
+            fi
+        language: system
+        verbose: true
+        stages: [commit-msg]
diff --git a/docker/sglang/Dockerfile b/docker/sglang/Dockerfile
new file mode 100644
index 000000000000..8490362401a0
--- /dev/null
+++ b/docker/sglang/Dockerfile
@@ -0,0 +1,98 @@
+# Declare the argument as default to use as input
+ARG BASE_IMAGE=lmsysorg/sglang:v0.5.5-cu129-amd64
+
+# Use input argument as base image
+FROM $BASE_IMAGE AS base
+
+# ====================================================
+# ====================== common ======================
+# ====================================================
+
+ARG PYTHON="python3"
+ARG EFA_VERSION="1.43.3"
+
+LABEL maintainer="Amazon AI"
+LABEL dlc_major_version="1"
+
+ENV DEBIAN_FRONTEND=noninteractive \
+  LANG=C.UTF-8 \
+  LC_ALL=C.UTF-8 \
+  DLC_CONTAINER_TYPE=general \
+  # Python won’t try to write .pyc or .pyo files on the import of source modules
+  # Force stdin, stdout and stderr to be totally unbuffered. Good for logging
+  PYTHONDONTWRITEBYTECODE=1 \
+  PYTHONUNBUFFERED=1 \
+  PYTHONIOENCODING=UTF-8 \
+  LD_LIBRARY_PATH="/usr/local/lib:/opt/amazon/ofi-nccl/lib/x86_64-linux-gnu:/opt/amazon/openmpi/lib:/opt/amazon/efa/lib:/usr/local/cuda/lib64:${LD_LIBRARY_PATH}" \
+  PATH="/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda/bin:${PATH}"
+
+WORKDIR /
+
+# Copy artifacts
+# ===============
+COPY ./scripts/telemetry/deep_learning_container.py /usr/local/bin/deep_learning_container.py
+COPY ./scripts/telemetry/bash_telemetry.sh /usr/local/bin/bash_telemetry.sh
+COPY ./scripts/common/install_efa.sh install_efa.sh
+COPY ./scripts/common/start_cuda_compat.sh /usr/local/bin/start_cuda_compat.sh
+
+RUN chmod +x /usr/local/bin/deep_learning_container.py \
+  && chmod +x /usr/local/bin/bash_telemetry.sh \
+  && chmod +x /usr/local/bin/start_cuda_compat.sh
+
+# Install EFA and remove vulnerable nvjpeg
+# =========================================
+RUN bash install_efa.sh ${EFA_VERSION} \
+  && rm install_efa.sh \
+  && mkdir -p /tmp/nvjpeg \
+  && cd /tmp/nvjpeg \
+  # latest cu12 libnvjpeg available is cu124
+  && wget https://developer.download.nvidia.com/compute/cuda/redist/libnvjpeg/linux-x86_64/libnvjpeg-linux-x86_64-12.4.0.76-archive.tar.xz \
+  && tar -xvf libnvjpeg-linux-x86_64-12.4.0.76-archive.tar.xz \
+  && rm -rf /usr/local/cuda/targets/x86_64-linux/lib/libnvjpeg* \
+  && rm -rf /usr/local/cuda/targets/x86_64-linux/include/nvjpeg.h \
+  && cp libnvjpeg-linux-x86_64-12.4.0.76-archive/lib/libnvjpeg* /usr/local/cuda/targets/x86_64-linux/lib/ \
+  && cp libnvjpeg-linux-x86_64-12.4.0.76-archive/include/* /usr/local/cuda/targets/x86_64-linux/include/ \
+  && rm -rf /tmp/nvjpeg \
+  # create symlink for python
+  && rm -rf /usr/bin/python \
+  && ln -s /usr/bin/python3 /usr/bin/python \
+  # remove cuobjdump and nvdisasm
+  && rm -rf /usr/local/cuda/bin/cuobjdump* \
+  && rm -rf /usr/local/cuda/bin/nvdisasm*
+
+# Run OSS compliance script
+# ==========================
+RUN echo 'source /usr/local/bin/bash_telemetry.sh' >> /etc/bash.bashrc \
+  # OSS compliance - use Python zipfile instead of unzip
+  && HOME_DIR=/root \
+  && curl -o ${HOME_DIR}/oss_compliance.zip https://aws-dlinfra-utilities.s3.amazonaws.com/oss_compliance.zip \
+  && python3 -c "import zipfile, os; zipfile.ZipFile('/root/oss_compliance.zip').extractall('/root/'); os.remove('/root/oss_compliance.zip')" \
+  && cp ${HOME_DIR}/oss_compliance/test/testOSSCompliance /usr/local/bin/testOSSCompliance \
+  && chmod +x /usr/local/bin/testOSSCompliance \
+  && chmod +x ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh \
+  && ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh ${HOME_DIR} ${PYTHON} \
+  # clean
+  && rm -rf ${HOME_DIR}/oss_compliance* \
+  && rm -rf /tmp/tmp* \
+  && rm -rf /tmp/uv* \
+  && rm -rf /var/lib/apt/lists/* \
+  && rm -rf /root/.cache | true
+
+# =======================================================
+# ====================== sagemaker ======================
+# =======================================================
+
+FROM base AS sglang-sagemaker
+
+ARG CACHE_REFRESH=0
+RUN dpkg -l | grep -E "cuda|nvidia|libnv" | awk '{print $2}' | xargs apt-mark hold \
+  && apt-get update \
+  && apt-get upgrade -y \
+  && apt-get clean
+
+RUN rm -rf /tmp/*
+
+COPY ./scripts/sglang/sagemaker_entrypoint.sh /usr/bin/serve
+RUN chmod +x /usr/bin/serve
+
+ENTRYPOINT ["/usr/bin/serve"]
\ No newline at end of file
diff --git a/scripts/common/start_cuda_compat.sh b/scripts/common/start_cuda_compat.sh
new file mode 100755
index 000000000000..791d355c5abe
--- /dev/null
+++ b/scripts/common/start_cuda_compat.sh
@@ -0,0 +1,25 @@
+#!/bin/bash
+
+verlte() {
+  [ "$1" = "$2" ] && return 1 || [ "$1" = "$(echo -e "$1\n$2" | sort -V | head -n1)" ]
+}
+
+COMPAT_FILE=/usr/local/cuda/compat/libcuda.so.1
+if [ -f $COMPAT_FILE ]; then
+  CUDA_COMPAT_MAX_DRIVER_VERSION=$(readlink $COMPAT_FILE | cut -d'.' -f 3-)
+  echo "CUDA compat package should be installed for NVIDIA driver smaller than ${CUDA_COMPAT_MAX_DRIVER_VERSION}"
+  NVIDIA_DRIVER_VERSION=$(sed -n 's/^NVRM.*Kernel Module *\([0-9.]*\).*$/\1/p' /proc/driver/nvidia/version 2>/dev/null || true)
+  if [ -z "$NVIDIA_DRIVER_VERSION" ]; then
+    NVIDIA_DRIVER_VERSION=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader --id=0 2>/dev/null || true)
+  fi
+  echo "Current installed NVIDIA driver version is ${NVIDIA_DRIVER_VERSION}"
+  if verlte $NVIDIA_DRIVER_VERSION $CUDA_COMPAT_MAX_DRIVER_VERSION; then
+    echo "Adding CUDA compat to LD_LIBRARY_PATH"
+    export LD_LIBRARY_PATH=/usr/local/cuda/compat:$LD_LIBRARY_PATH
+    echo $LD_LIBRARY_PATH
+  else
+    echo "Skipping CUDA compat setup as newer NVIDIA driver is installed"
+  fi
+else
+  echo "Skipping CUDA compat setup as package not found"
+fi
diff --git a/scripts/sglang/sagemaker_entrypoint.sh b/scripts/sglang/sagemaker_entrypoint.sh
new file mode 100755
index 000000000000..0f13cf5cb002
--- /dev/null
+++ b/scripts/sglang/sagemaker_entrypoint.sh
@@ -0,0 +1,42 @@
+#!/bin/bash
+# Check if telemetry file exists before executing
+# Execute telemetry script if it exists, suppress errors
+bash /usr/local/bin/bash_telemetry.sh >/dev/null 2>&1 || true
+
+if command -v nvidia-smi >/dev/null 2>&1 && command -v nvcc >/dev/null 2>&1; then
+    bash /usr/local/bin/start_cuda_compat.sh
+fi
+
+echo "Starting server"
+
+PREFIX="SM_SGLANG_"
+ARG_PREFIX="--"
+
+ARGS=()
+
+while IFS='=' read -r key value; do
+    arg_name=$(echo "${key#"${PREFIX}"}" | tr '[:upper:]' '[:lower:]' | tr '_' '-')
+
+    ARGS+=("${ARG_PREFIX}${arg_name}")
+    if [ -n "$value" ]; then
+        ARGS+=("$value")
+    fi
+done < <(env | grep "^${PREFIX}")
+
+# Add default port only if not already set
+if ! [[ " ${ARGS[@]} " =~ " --port " ]]; then
+    ARGS+=(--port "${SM_SGLANG_PORT:-8080}")
+fi
+
+# Add default host only if not already set
+if ! [[ " ${ARGS[@]} " =~ " --host " ]]; then
+    ARGS+=(--host "${SM_SGLANG_HOST:-0.0.0.0}")
+fi
+
+# Add default model-path only if not already set
+if ! [[ " ${ARGS[@]} " =~ " --model-path " ]]; then
+    ARGS+=(--model-path "${SM_SGLANG_MODEL_PATH:-/opt/ml/model}")
+fi
+
+echo "Running command: exec python3 -m sglang.launch_server ${ARGS[@]}"
+exec python3 -m sglang.launch_server "${ARGS[@]}"
diff --git a/test/__init__.py b/test/__init__.py
new file mode 100644
index 000000000000..6181bd5dfc31
--- /dev/null
+++ b/test/__init__.py
@@ -0,0 +1,26 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+
+import logging
+import sys
+
+from test_utils.logger import ColoredFormatter
+
+logger = logging.getLogger(__name__)
+logger.setLevel(logging.DEBUG)
+
+console_handler = logging.StreamHandler(sys.stdout)
+console_handler.setFormatter(ColoredFormatter())
+console_handler.setLevel(logging.DEBUG)
+
+logger.addHandler(console_handler)
diff --git a/test/conftest.py b/test/conftest.py
new file mode 100644
index 000000000000..0ac31d0f8c00
--- /dev/null
+++ b/test/conftest.py
@@ -0,0 +1,39 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+"""Common pytest fixtures for all tests under module test/"""
+
+import pytest
+from test_utils.aws import AWSSessionManager
+from test_utils.constants import DEFAULT_REGION
+
+
+def pytest_addoption(parser):
+    parser.addoption("--image-uri", action="store", help="Image URI to be tested")
+    parser.addoption(
+        "--region", action="store", default=DEFAULT_REGION, help="AWS Region to test image on AWS"
+    )
+
+
+@pytest.fixture(scope="session")
+def image_uri(request):
+    return request.config.getoption("--image-uri")
+
+
+@pytest.fixture(scope="session")
+def region(request):
+    return request.config.getoption("--region")
+
+
+@pytest.fixture(scope="session")
+def aws_session(region):
+    return AWSSessionManager(region)
diff --git a/test/requirements.txt b/test/requirements.txt
new file mode 100644
index 000000000000..1fcff467b343
--- /dev/null
+++ b/test/requirements.txt
@@ -0,0 +1,3 @@
+boto3
+botocore
+pytest
diff --git a/test/sglang/sagemaker/requirements.txt b/test/sglang/sagemaker/requirements.txt
new file mode 100644
index 000000000000..30f66577a3b2
--- /dev/null
+++ b/test/sglang/sagemaker/requirements.txt
@@ -0,0 +1 @@
+sagemaker
diff --git a/test/sglang/sagemaker/test_sm_endpoint.py b/test/sglang/sagemaker/test_sm_endpoint.py
new file mode 100644
index 000000000000..7e5de37a3507
--- /dev/null
+++ b/test/sglang/sagemaker/test_sm_endpoint.py
@@ -0,0 +1,170 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+"""Integration test for serving endpoint with SGLang DLC"""
+
+import json
+import logging
+from pprint import pformat
+
+import pytest
+from botocore.exceptions import ClientError
+from sagemaker import serializers
+from sagemaker.model import Model
+from sagemaker.predictor import Predictor
+from test_utils import clean_string, random_suffix_name, wait_for_status
+
+# To enable debugging, change logging.INFO to logging.DEBUG
+LOGGER = logging.getLogger(__name__)
+LOGGER.setLevel(logging.INFO)
+
+ENDPOINT_WAIT_PERIOD = 60
+ENDPOINT_WAIT_LENGTH = 30
+ENDPOINT_INSERVICE = "InService"
+
+
+def get_endpoint_status(sagemaker_client, endpoint_name):
+    response = sagemaker_client.describe_endpoint(EndpointName=endpoint_name)
+    LOGGER.debug(f"Describe endpoint response: {pformat(response)}")
+    return response["EndpointStatus"]
+
+
+def get_hf_token(aws_session):
+    LOGGER.info("Retrieving HuggingFace token from AWS Secrets Manager...")
+    token_path = "test/hf_token"
+
+    try:
+        get_secret_value_response = aws_session.secretsmanager.get_secret_value(SecretId=token_path)
+        LOGGER.info("Successfully retrieved HuggingFace token")
+    except ClientError as e:
+        LOGGER.error(f"Failed to retrieve HuggingFace token: {e}")
+        raise e
+
+    # Do not print secrets token in logs
+    response = json.loads(get_secret_value_response["SecretString"])
+    token = response.get("HF_TOKEN")
+    return token
+
+
+@pytest.fixture(scope="function")
+def model_id(request):
+    # Return the model_id given by the test parameter
+    return request.param
+
+
+@pytest.fixture(scope="function")
+def instance_type(request):
+    # Return the model_id given by the test parameter
+    return request.param
+
+
+@pytest.fixture(scope="function")
+def model_package(aws_session, image_uri, model_id):
+    sagemaker_client = aws_session.sagemaker
+    sagemaker_role = aws_session.iam_resource.Role("SageMakerRole").arn
+    cleaned_id = clean_string(model_id.split("/")[1], "_./")
+    model_name = random_suffix_name(f"sglang-{cleaned_id}-model-package", 50)
+
+    LOGGER.debug(f"Using image: {image_uri}")
+    LOGGER.debug(f"Model ID: {model_id}")
+
+    LOGGER.info(f"Creating SageMaker model: {model_name}...")
+    hf_token = get_hf_token(aws_session)
+    model = Model(
+        name=model_name,
+        image_uri=image_uri,
+        role=sagemaker_role,
+        predictor_cls=Predictor,
+        env={
+            "SM_SGLANG_MODEL_PATH": model_id,
+            "HF_TOKEN": hf_token,
+        },
+    )
+    LOGGER.info("Model created successfully")
+
+    yield model
+
+    LOGGER.info(f"Deleting model: {model_name}")
+    sagemaker_client.delete_model(ModelName=model_name)
+
+
+@pytest.fixture(scope="function")
+def model_endpoint(aws_session, model_package, instance_type):
+    sagemaker_client = aws_session.sagemaker
+    model = model_package
+    cleaned_instance = clean_string(instance_type, "_./")
+    endpoint_name = random_suffix_name(f"sglang-{cleaned_instance}-endpoint", 50)
+
+    LOGGER.debug(f"Using instance type: {instance_type}")
+
+    LOGGER.info("Starting endpoint deployment (this may take 10-15 minutes)...")
+    predictor = model.deploy(
+        instance_type=instance_type,
+        initial_instance_count=1,
+        endpoint_name=endpoint_name,
+        inference_ami_version="al2-ami-sagemaker-inference-gpu-3-1",
+        serializer=serializers.JSONSerializer(),
+        wait=True,
+    )
+    LOGGER.info("Endpoint deployment completed successfully")
+
+    LOGGER.info(f"Waiting for endpoint {ENDPOINT_INSERVICE} status ...")
+    assert wait_for_status(
+        ENDPOINT_INSERVICE,
+        ENDPOINT_WAIT_PERIOD,
+        ENDPOINT_WAIT_LENGTH,
+        get_endpoint_status,
+        sagemaker_client,
+        endpoint_name,
+    )
+
+    yield predictor
+
+    LOGGER.info(f"Deleting endpoint: {endpoint_name}")
+    sagemaker_client.delete_endpoint(EndpointName=endpoint_name)
+
+    LOGGER.info(f"Deleting endpoint configuration: {endpoint_name}")
+    sagemaker_client.delete_endpoint_config(EndpointConfigName=endpoint_name)
+
+
+@pytest.mark.parametrize("instance_type", ["ml.g5.12xlarge"], indirect=True)
+@pytest.mark.parametrize("model_id", ["Qwen/Qwen3-0.6B"], indirect=True)
+def test_sglang_sagemaker_endpoint(model_endpoint, model_id):
+    predictor = model_endpoint
+
+    prompt = "Write a python script to calculate square of n"
+    payload = {
+        "model": model_id,
+        "messages": [{"role": "user", "content": prompt}],
+        "max_tokens": 2400,
+        "temperature": 0.01,
+        "top_p": 0.9,
+        "top_k": 50,
+    }
+    LOGGER.debug(f"Sending inference request with payload: {pformat(payload)}")
+
+    response = predictor.predict(payload)
+    LOGGER.info("Inference request invoked successfully")
+
+    if isinstance(response, bytes):
+        response = response.decode("utf-8")
+
+    if isinstance(response, str):
+        try:
+            response = json.loads(response)
+        except json.JSONDecodeError:
+            LOGGER.warning("Response is not valid JSON. Returning as string.")
+
+    assert response, "Model response is empty, failing endpoint test!"
+
+    LOGGER.info(f"Model response: {pformat(response)}")
+    LOGGER.info("Inference test successful!")
diff --git a/test/test_utils/__init__.py b/test/test_utils/__init__.py
new file mode 100644
index 000000000000..7c307b1b8259
--- /dev/null
+++ b/test/test_utils/__init__.py
@@ -0,0 +1,60 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+"""Common utility functions for all tests under module test/
+For test utility functions, please appropriately declare function argument types
+and their output types for readability and reusability.
+When necessary, use docstrings to explain the functions' mechanisms.
+"""
+
+import logging
+import random
+import string
+import time
+from collections.abc import Callable
+from typing import Any
+
+LOGGER = logging.getLogger(__name__)
+LOGGER.setLevel(logging.INFO)
+
+
+def random_suffix_name(resource_name: str, max_length: int, delimiter: str = "-") -> str:
+    rand_length = max_length - len(resource_name) - len(delimiter)
+    rand = "".join(
+        random.choice(string.ascii_lowercase + string.digits) for _ in range(rand_length)
+    )
+    return f"{resource_name}{delimiter}{rand}"
+
+
+def clean_string(text: str, symbols_to_remove: str, replacement: str = "-") -> str:
+    for symbol in symbols_to_remove:
+        text = text.replace(symbol, replacement)
+    return text
+
+
+def wait_for_status(
+    expected_status: str,
+    wait_periods: int,
+    period_length: int,
+    get_status_method: Callable[[Any], str],
+    *method_args: Any,
+) -> bool:
+    actual_status = None
+    for i in range(wait_periods):
+        time.sleep(period_length)
+        LOGGER.debug(f"Time passed while waiting: {period_length * (i + 1)}s.")
+        actual_status = get_status_method(*method_args)
+        if actual_status == expected_status:
+            return True
+
+    LOGGER.error(f"Wait for status: {expected_status} timed out. Actual status: {actual_status}")
+    return False
diff --git a/test/test_utils/aws.py b/test/test_utils/aws.py
new file mode 100644
index 000000000000..a6e1c49dbed9
--- /dev/null
+++ b/test/test_utils/aws.py
@@ -0,0 +1,42 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+"""AWS Session Manager for all AWS boto3 API resources"""
+
+import boto3
+from test_utils.constants import DEFAULT_REGION
+
+
+class AWSSessionManager:
+    def __init__(self, region=DEFAULT_REGION, profile_name=None):
+        if profile_name:
+            self.session = boto3.Session(profile_name=profile_name, region_name=region)
+        else:
+            self.session = boto3.Session(region_name=region)
+
+        # Client API
+        self.cloudwatch = self.session.client("cloudwatch")
+        self.codebuild = self.session.client("codebuild")
+        self.codepipeline = self.session.client("codepipeline")
+        self.ec2 = self.session.client("ec2")
+        self.events = self.session.client("events")
+        self.iam = self.session.client("iam")
+        self.resource_groups = self.session.client("resource-groups")
+        self.scheduler = self.session.client("scheduler")
+        self.secretsmanager = self.session.client("secretsmanager")
+        self.sts = self.session.client("sts")
+        self.s3 = self.session.client("s3")
+        self.sagemaker = self.session.client("sagemaker")
+
+        # Resource API
+        self.iam_resource = self.session.resource("iam")
+        self.s3_resource = self.session.resource("s3")
diff --git a/test/test_utils/constants.py b/test/test_utils/constants.py
new file mode 100644
index 000000000000..eaa7babdfaf9
--- /dev/null
+++ b/test/test_utils/constants.py
@@ -0,0 +1 @@
+DEFAULT_REGION = "us-west-2"
diff --git a/test/test_utils/logger.py b/test/test_utils/logger.py
new file mode 100644
index 000000000000..aa9725085c73
--- /dev/null
+++ b/test/test_utils/logger.py
@@ -0,0 +1,41 @@
+# Copyright 2018-2020 Amazon.com, Inc. or its affiliates. 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. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file 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.
+"""Logging handler for nice formatted logs"""
+
+import logging
+
+
+# Custom formatter
+class ColoredFormatter(logging.Formatter):
+    COLORS = {
+        "DEBUG": {"text": "\x1b[36;20m", "bold": "\x1b[1;36;20m", "underline": "\x1b[4;36;20m"},
+        "INFO": {"text": "\x1b[38;20m", "bold": "\x1b[1;38;20m", "underline": "\x1b[4;38;20m"},
+        "WARNING": {"text": "\x1b[33;20m", "bold": "\x1b[1;33;20m", "underline": "\x1b[4;33;20m"},
+        "ERROR": {"text": "\x1b[31;20m", "bold": "\x1b[1;31;20m", "underline": "\x1b[4;31;20m"},
+        "CRITICAL": {"text": "\x1b[31;1m", "bold": "\x1b[1;31;1m", "underline": "\x1b[4;31;1m"},
+        "RESET": "\x1b[0m",
+    }
+
+    def format(self, record):
+        colors = self.COLORS.get(record.levelname, self.COLORS["DEBUG"])
+        reset = self.COLORS["RESET"]
+
+        # Create formatted string with different styles
+        format_str = (
+            f"{colors['bold']}%(asctime)s{reset} - "
+            f"{colors['text']}%(name)s{reset} - "
+            f"{colors['underline']}%(levelname)s{reset} - "
+            f"{colors['text']}%(message)s{reset}"
+        )
+        formatter = logging.Formatter(format_str)
+        return formatter.format(record)