diff --git a/build.py b/build.py
index 2a91e4b570..88190ba484 100755
--- a/build.py
+++ b/build.py
@@ -819,7 +819,21 @@ def fastertransformer_cmake_args():
 
 
 def tensorrtllm_cmake_args(images):
-    cargs = []
+    cmake_script.cmd("apt-get update && apt-get install -y libcudnn8-dev && ldconfig")
+    cmake_script.cmd(
+        "python3 ../tensorrt_llm/scripts/build_wheel.py --trt_root /usr/local/tensorrt -i -c -j 18"
+    )
+    cargs = [
+        cmake_backend_arg(
+            "tensorrtllm",
+            "TRT_LIB_DIR",
+            None,
+            "${TRT_ROOT}/targets/${ARCH}-linux-gnu/lib",
+        ),
+        cmake_backend_arg(
+            "tensorrtllm", "TRT_INCLUDE_DIR", None, "${TRT_ROOT}/include"
+        ),
+    ]
     cargs.append(cmake_backend_enable("tensorrtllm", "USE_CXX11_ABI", True))
     return cargs
 
@@ -1094,10 +1108,19 @@ def create_dockerfile_linux(
 RUN python3 -m pip install --upgrade pip \\
       && pip3 install transformers
 
+# Drop the static libs
+RUN ARCH="$(uname -i)" \\
+      && rm -f ${TRT_ROOT}/targets/${ARCH}-linux-gnu/lib/libnvinfer*.a \\
+          ${TRT_ROOT}/targets/${ARCH}-linux-gnu/lib/libnvonnxparser_*.a
+
 # Install TensorRT-LLM
+RUN python3 -m pip install /opt/tritonserver/backends/tensorrtllm/tensorrt_llm-*.whl -U --pre --extra-index-url https://pypi.nvidia.com \\
+        && rm -fv /opt/tritonserver/backends/tensorrtllm/tensorrt_llm-*.whl
 RUN find /usr -name libtensorrt_llm.so -exec dirname {} \; > /etc/ld.so.conf.d/tensorrt-llm.conf
 RUN find /opt/tritonserver -name libtritonserver.so -exec dirname {} \; > /etc/ld.so.conf.d/triton-tensorrtllm-worker.conf
 
+RUN pip3 install setuptools==69.5.1 grpcio-tools==1.64.0
+
 ENV LD_LIBRARY_PATH=/usr/local/tensorrt/lib/:/opt/tritonserver/backends/tensorrtllm:$LD_LIBRARY_PATH
 """
     with open(os.path.join(ddir, dockerfile_name), "w") as dfile:
@@ -1702,6 +1725,11 @@ def tensorrtllm_postbuild(cmake_script, repo_install_dir, tensorrtllm_be_dir):
     # TODO: Update the CMakeLists.txt of TRT-LLM backend to install the artifacts to the correct location
     cmake_destination_dir = os.path.join(repo_install_dir, "backends/tensorrtllm")
     cmake_script.mkdir(cmake_destination_dir)
+    # Copy over the TRT-LLM wheel for later installation
+    cmake_script.cp(
+        os.path.join(tensorrtllm_be_dir, "tensorrt_llm", "build", "tensorrt_llm-*.whl"),
+        cmake_destination_dir,
+    )
 
     # Copy over the TRT-LLM backend libraries
     cmake_script.cp(