diff --git a/CMakeLists.txt b/CMakeLists.txt
index 36cd7d8..f760070 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -1,17 +1,34 @@
 cmake_minimum_required(VERSION 3.9)
-project(MyProject)
+project(MyProject LANGUAGES CUDA CXX)
+
+add_compile_options(-Wall -Wextra -Wpedantic)
 
 set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CUDA_ARCHITECTURES 61)
+set(CUDA_SEPARABLE_COMPILATION ON)
+
 
-set(SOURCE_FILES main.cpp)
-add_executable(${CMAKE_PROJECT_NAME}_run ${SOURCE_FILES})
 
 include_directories(src)
+include_directories(kernels)
+include_directories(/usr/local/cuda-12.8/include)
+
 
 add_subdirectory(src)
+add_subdirectory(kernels)
 add_subdirectory(tests)
 
-target_link_libraries(${CMAKE_PROJECT_NAME}_run ${CMAKE_PROJECT_NAME}_lib)
+add_executable(${CMAKE_PROJECT_NAME}_run main.cpp)
+
+
+
+target_link_libraries(
+    ${CMAKE_PROJECT_NAME}_run 
+    PRIVATE
+    ${CMAKE_PROJECT_NAME}_lib 
+    ${CMAKE_PROJECT_NAME}_cuda_lib 
+    ${CUDA_LIBRARIES}
+)
 
 # Doxygen Build
 option(BUILD_DOC "Build Documentation" ON)
@@ -35,4 +52,4 @@ if(DOXYGEN_FOUND)
             VERBATIM)
 else(DOXYGEN_FOUND)
     message("Doxygen needs to be installed to generate the documentation.")
-endif(DOXYGEN_FOUND)
\ No newline at end of file
+endif(DOXYGEN_FOUND)
diff --git a/kernels/CMakeLists.txt b/kernels/CMakeLists.txt
new file mode 100644
index 0000000..f9a1174
--- /dev/null
+++ b/kernels/CMakeLists.txt
@@ -0,0 +1,18 @@
+project(${CMAKE_PROJECT_NAME}_cuda_lib CUDA CXX)
+
+set(HEADER_FILES
+    hello_world.h
+)
+set(SOURCE_FILES
+    hello_world.cu
+)
+
+# The library contains header and source files.
+add_library(${CMAKE_PROJECT_NAME}_cuda_lib STATIC
+    ${SOURCE_FILES}
+    ${HEADER_FILES}
+)
+
+if(CMAKE_COMPILER_IS_GNUCXX)
+  target_compile_options(${CMAKE_PROJECT_NAME}_cuda_lib PRIVATE -Wno-gnu-line-marker)
+endif()
diff --git a/kernels/hello_world.cu b/kernels/hello_world.cu
new file mode 100644
index 0000000..7c65115
--- /dev/null
+++ b/kernels/hello_world.cu
@@ -0,0 +1,46 @@
+#include <cuda_runtime.h>
+#include <stdio.h>
+
+__global__ void hello_cuda() {
+  printf("Hello CUDA from thread %d\n", threadIdx.x);
+}
+
+extern "C" void launch_hello_cuda() {
+  // First check device properties
+  cudaDeviceProp prop;
+  cudaGetDeviceProperties(&prop, 1);
+  printf("Using device: %s with compute capability %d.%d\n", prop.name,
+         prop.major, prop.minor);
+
+  hello_cuda<<<1, 10>>>();
+  cudaDeviceSynchronize();
+  fflush(stdout);
+}
+
+extern "C" void check_cuda() {
+  int deviceCount = 0;
+  cudaError_t error = cudaGetDeviceCount(&deviceCount);
+
+  if (error != cudaSuccess) {
+    printf("CUDA error: %s\n", cudaGetErrorString(error));
+  }
+
+  printf("Found %d CUDA devices\n", deviceCount);
+
+  for (int i = 0; i < deviceCount; i++) {
+    cudaDeviceProp prop;
+    cudaGetDeviceProperties(&prop, i);
+
+    printf("Device %d: %s\n", i, prop.name);
+    printf("  Compute capability: %d.%d\n", prop.major, prop.minor);
+    printf("  Total global memory: %.2f GB\n",
+           static_cast<float>(prop.totalGlobalMem) / (1024 * 1024 * 1024));
+    printf("  Multiprocessors: %d\n", prop.multiProcessorCount);
+    printf("  Max threads per block: %d\n", prop.maxThreadsPerBlock);
+    printf("  Max threads dimensions: (%d, %d, %d)\n", prop.maxThreadsDim[0],
+           prop.maxThreadsDim[1], prop.maxThreadsDim[2]);
+    printf("  Max grid dimensions: (%d, %d, %d)\n", prop.maxGridSize[0],
+           prop.maxGridSize[1], prop.maxGridSize[2]);
+    printf("\n");
+  }
+}
diff --git a/kernels/hello_world.h b/kernels/hello_world.h
new file mode 100644
index 0000000..4024e2e
--- /dev/null
+++ b/kernels/hello_world.h
@@ -0,0 +1,10 @@
+#ifndef HELLO_WORLD_CU_H
+#define HELLO_WORLD_CU_H
+
+extern "C" {
+// Declaration of the CUDA function that will be called from C++
+void launch_hello_cuda();
+void check_cuda();
+}
+
+#endif // HELLO_WORLD_CU_H
diff --git a/main.cpp b/main.cpp
index d269c89..e3c8734 100644
--- a/main.cpp
+++ b/main.cpp
@@ -1,3 +1,10 @@
+#include "hello_world.h"
+#include <iostream>
+
 int main() {
+  std::cout << "Starting CUDA example..." << std::endl; // Using endl to flush
+  check_cuda();
+  launch_hello_cuda();
+  std::cout << "Ending CUDA example" << std::endl; // Using endl to flush
   return 0;
-}
\ No newline at end of file
+}
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 94717aa..6c6cd8f 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -1,17 +1,16 @@
-project(${CMAKE_PROJECT_NAME}_lib)
+project(${CMAKE_PROJECT_NAME}_lib CUDA CXX)
 
 set(HEADER_FILES
+    ./test.h
 )
 set(SOURCE_FILES
+    ./test.cpp
 )
 
-if (EXISTS ${SOURCE_FILES})
-    # The library contains header and source files.
-    add_library(${CMAKE_PROJECT_NAME}_lib STATIC
-        ${SOURCE_FILES}
-        ${HEADER_FILES}
-    )
-else()
-    # The library only contains header files.
-    add_library(${CMAKE_PROJECT_NAME}_lib INTERFACE)
-endif()
\ No newline at end of file
+# The library contains header and source files.
+add_library(${CMAKE_PROJECT_NAME}_lib
+    ${SOURCE_FILES}
+    ${HEADER_FILES}
+)
+
+target_include_directories(${CMAKE_PROJECT_NAME}_lib PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})
diff --git a/src/test.cpp b/src/test.cpp
new file mode 100644
index 0000000..eec7934
--- /dev/null
+++ b/src/test.cpp
@@ -0,0 +1,4 @@
+#include "test.h"
+#include <iostream>
+
+void test_hello() { std::cout << "Hello!"; }
diff --git a/src/test.h b/src/test.h
new file mode 100644
index 0000000..b9c7ab1
--- /dev/null
+++ b/src/test.h
@@ -0,0 +1,2 @@
+#include <iostream>
+void test_hello();