diff --git a/.gitignore b/.gitignore
index c98f56c972bc..f0e194758975 100644
--- a/.gitignore
+++ b/.gitignore
@@ -61,3 +61,12 @@ mshadow
 data
 recommonmark
 bin
+
+# msvc
+*.pdb
+*.sdf
+*.opensdf
+*.deps
+*.cache
+*.exp
+mxnet.egg-info/
\ No newline at end of file
diff --git a/example/mnist/data.py b/example/mnist/data.py
index 38f71263ea3f..f533ada0c859 100644
--- a/example/mnist/data.py
+++ b/example/mnist/data.py
@@ -1,8 +1,10 @@
 # pylint: skip-file
 """ data iterator for mnist """
 import sys
-sys.path.insert(0, "../../python/")
-sys.path.append("../../tests/python/common")
+import os
+curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
+sys.path.insert(0, os.path.join(curr_path, "../../python/"))
+sys.path.append(os.path.join(curr_path, "../../tests/python/common"))
 import get_data
 import mxnet as mx
 
diff --git a/python/mxnet/libinfo.py b/python/mxnet/libinfo.py
index 821d33e5bce2..030f2eb7436e 100644
--- a/python/mxnet/libinfo.py
+++ b/python/mxnet/libinfo.py
@@ -16,10 +16,11 @@ def find_lib_path():
     api_path = os.path.join(curr_path, '../../lib/')
     dll_path = [curr_path, api_path]
     if os.name == 'nt':
+        vs_configuration = 'Release'
         if platform.architecture()[0] == '64bit':
-            dll_path.append(os.path.join(api_path, '../../windows/x64/Release/'))
+            dll_path.append(os.path.join(curr_path, '../../windows/x64', vs_configuration))
         else:
-            dll_path.append(os.path.join(api_path, '../../windows/Release/'))
+            dll_path.append(os.path.join(curr_path, '../../windows', vs_configuration))
     if os.name == 'nt':
         dll_path = [os.path.join(p, 'mxnet.dll') for p in dll_path]
     else:
diff --git a/src/common/object_pool.h b/src/common/object_pool.h
index 0306ed6ef355..5787f53ff497 100644
--- a/src/common/object_pool.h
+++ b/src/common/object_pool.h
@@ -157,8 +157,13 @@ void ObjectPool<T>::AllocateChunk() {
   static_assert(alignof(LinkedList) % alignof(T) == 0, "ObjectPooll Invariant");
   static_assert(kPageSize % alignof(LinkedList) == 0, "ObjectPooll Invariant");
   void* new_chunk_ptr;
+#ifdef _MSC_VER
+  new_chunk_ptr = _aligned_malloc(kPageSize, kPageSize);
+  CHECK_NE(new_chunk_ptr, NULL) << "Allocation failed";
+#else
   int ret = posix_memalign(&new_chunk_ptr, kPageSize, kPageSize);
   CHECK_EQ(ret, 0) << "Allocation failed";
+#endif
   allocated_.emplace_back(new_chunk_ptr);
   auto new_chunk = static_cast<LinkedList*>(new_chunk_ptr);
   auto size = kPageSize / sizeof(LinkedList);
diff --git a/src/io/image_augmenter.h b/src/io/image_augmenter.h
index cd50c5e10b08..b76b8da4650e 100644
--- a/src/io/image_augmenter.h
+++ b/src/io/image_augmenter.h
@@ -114,6 +114,9 @@ class ImageAugmenter {
     }
   }
 #if MXNET_USE_OPENCV
+#ifdef _MSC_VER
+#define M_PI CV_PI
+#endif
   /*!
    * \brief augment src image, store result into dst
    *   this function is not thread safe, and will only be called by one thread
diff --git a/src/io/iter_mnist.cc b/src/io/iter_mnist.cc
index 89fb7efb026f..7ba717b1c7f9 100644
--- a/src/io/iter_mnist.cc
+++ b/src/io/iter_mnist.cc
@@ -171,7 +171,11 @@ class MNISTIter: public IIterator<TBlobBatch> {
     unsigned char buf[4];
     CHECK(fi->Read(buf, sizeof(buf)) == sizeof(buf))
         << "invalid mnist format";
+#ifdef _MSC_VER
+    return (buf[0] << 24 | buf[1] << 16 | buf[2] << 8 | buf[3]);
+#else
     return reinterpret_cast<int>(buf[0] << 24 | buf[1] << 16 | buf[2] << 8 | buf[3]);
+#endif
   }
 
  private:
diff --git a/src/storage/cpu_device_storage.h b/src/storage/cpu_device_storage.h
index 82aa0e7573f6..6838af037535 100644
--- a/src/storage/cpu_device_storage.h
+++ b/src/storage/cpu_device_storage.h
@@ -39,7 +39,9 @@ class CPUDeviceStorage {
 
 inline void* CPUDeviceStorage::Alloc(size_t size) {
 #if _MSC_VER
-  return CHECK_NOTNULL(_aligned_malloc(size, alignment_));
+  void* ptr;
+  ptr = _aligned_malloc(size, alignment_);
+  return CHECK_NOTNULL(ptr);
 #else
   void* ptr;
   int ret = posix_memalign(&ptr, alignment_, size);
@@ -48,7 +50,13 @@ inline void* CPUDeviceStorage::Alloc(size_t size) {
 #endif
 }
 
-inline void CPUDeviceStorage::Free(void* ptr) { free(ptr); }
+inline void CPUDeviceStorage::Free(void* ptr) {
+#if _MSC_VER
+  _aligned_free(ptr);
+#else
+  free(ptr);
+#endif
+}
 
 }  // namespace storage
 }  // namespace mxnet
diff --git a/windows/3rdparty/.gitignore b/windows/3rdparty/.gitignore
new file mode 100644
index 000000000000..5e7d2734cfc6
--- /dev/null
+++ b/windows/3rdparty/.gitignore
@@ -0,0 +1,4 @@
+# Ignore everything in this directory
+*
+# Except this file
+!.gitignore
diff --git a/windows/README.md b/windows/README.md
new file mode 100644
index 000000000000..f9a2ec4d189a
--- /dev/null
+++ b/windows/README.md
@@ -0,0 +1,32 @@
+Build in Visual Studio 2013
+=================
+
+Prepare
+---------
+
+1. Download 3rdparty libraries form [BaiduYun](http://pan.baidu.com/s/1x7fw6) or [OneDrive](Not available now),  and extract the files into `3rdparty`
+
+2. Download and install [Nov 2013 CTP](http://www.microsoft.com/en-us/download/details.aspx?id=41151), [PTVS](https://github.com/Microsoft/PTVS/releases).
+
+3. Copy all files in `C:\Program Files (x86)\Microsoft Visual C++ Compiler Nov 2013 CTP` to `C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC` and overwrite all existed files (of course, backup them before copying).
+
+Build
+----------
+1. Open mxnet.sln.
+
+2. Switch the compile mode to Release and x64.
+
+3. If you have MKL, please modify the include/library configs and set three preprocess macro to:
+  - MSHADOW_USE_CBLAS=0
+  - MSHADOW_USE_MKL=1
+  - USE_STATIC_MKL=1
+
+4. Modify the cuda device compute capability defined in the settings (`mxnet properties` -> `CUDA C/C++` -> `Device` -> `Code Generation`) to your GPU's compute capability (such as compute_30,sm_30). You can look up for your GPU's compute capability in https://en.wikipedia.org/wiki/CUDA . Some general GPUs' compute capabilities are listed below.
+
+5. Compile.
+
+| GPU                                         | Compute Capability    |
+| ------------------------------------------- |:---------------------:|
+| GTX660, 680, 760, 770                       | compute_30,sm_30      |
+| GTX780, Titan Z, Titan Black, K20, K40      | compute_35,sm_35      |
+| GTX960, 980, Titan X                        | compute_52,sm_52      |
diff --git a/windows/im2rec/im2rec.vcxproj b/windows/im2rec/im2rec.vcxproj
new file mode 100644
index 000000000000..7a101f329319
--- /dev/null
+++ b/windows/im2rec/im2rec.vcxproj
@@ -0,0 +1,162 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\tools\im2rec.cc" />
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{CF267D06-BE5D-4303-A013-33BBC076809F}</ProjectGuid>
+    <RootNamespace>im2rec</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+    <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 7.0.props" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;WIN64;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+    <CudaCompile>
+      <TargetMachinePlatform>64</TargetMachinePlatform>
+    </CudaCompile>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;WIN64;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>../../dmlc-core/include;../../mshadow;../../include;D:\deepLearning\caffe-windows\3rdparty\include;%(AdditionalIncludeDirectories);$(CudaToolkitIncludeDir)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>dmlc.lib;mxnet.lib;opencv_world300.lib;cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>D:\deepLearning\caffe-windows\3rdparty\lib;..\..\windows\x64\Release;%(AdditionalLibraryDirectories);$(CudaToolkitLibDir)</AdditionalLibraryDirectories>
+    </Link>
+    <PostBuildEvent>
+      <Command>
+      </Command>
+    </PostBuildEvent>
+    <CudaCompile>
+      <TargetMachinePlatform>64</TargetMachinePlatform>
+    </CudaCompile>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+    <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 7.0.targets" />
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/windows/mxnet.sln b/windows/mxnet.sln
new file mode 100644
index 000000000000..72d8948111de
--- /dev/null
+++ b/windows/mxnet.sln
@@ -0,0 +1,80 @@
+
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 2013
+VisualStudioVersion = 12.0.21005.1
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "mxnet", "mxnet\mxnet.vcxproj", "{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "dmlc", "..\dmlc-core\windows\dmlc\dmlc.vcxproj", "{6E6DDA36-69BA-4555-BA88-10166B11F2D2}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "im2rec", "im2rec\im2rec.vcxproj", "{CF267D06-BE5D-4303-A013-33BBC076809F}"
+EndProject
+Project("{888888A0-9F3D-457C-B088-3A5042F75D52}") = "python", "python\python.pyproj", "{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}"
+EndProject
+Global
+	GlobalSection(SolutionConfigurationPlatforms) = preSolution
+		Debug|Any CPU = Debug|Any CPU
+		Debug|Mixed Platforms = Debug|Mixed Platforms
+		Debug|Win32 = Debug|Win32
+		Debug|x64 = Debug|x64
+		Release|Any CPU = Release|Any CPU
+		Release|Mixed Platforms = Release|Mixed Platforms
+		Release|Win32 = Release|Win32
+		Release|x64 = Release|x64
+	EndGlobalSection
+	GlobalSection(ProjectConfigurationPlatforms) = postSolution
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|Any CPU.ActiveCfg = Debug|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|Mixed Platforms.ActiveCfg = Debug|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|Mixed Platforms.Build.0 = Debug|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|Win32.ActiveCfg = Debug|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|Win32.Build.0 = Debug|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|x64.ActiveCfg = Debug|x64
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Debug|x64.Build.0 = Debug|x64
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|Any CPU.ActiveCfg = Release|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|Mixed Platforms.ActiveCfg = Release|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|Mixed Platforms.Build.0 = Release|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|Win32.ActiveCfg = Release|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|Win32.Build.0 = Release|Win32
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|x64.ActiveCfg = Release|x64
+		{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}.Release|x64.Build.0 = Release|x64
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|Any CPU.ActiveCfg = Debug|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|Mixed Platforms.ActiveCfg = Debug|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|Mixed Platforms.Build.0 = Debug|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|Win32.ActiveCfg = Debug|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|Win32.Build.0 = Debug|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|x64.ActiveCfg = Debug|x64
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Debug|x64.Build.0 = Debug|x64
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|Any CPU.ActiveCfg = Release|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|Mixed Platforms.ActiveCfg = Release|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|Mixed Platforms.Build.0 = Release|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|Win32.ActiveCfg = Release|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|Win32.Build.0 = Release|Win32
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|x64.ActiveCfg = Release|x64
+		{6E6DDA36-69BA-4555-BA88-10166B11F2D2}.Release|x64.Build.0 = Release|x64
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|Any CPU.ActiveCfg = Debug|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|Mixed Platforms.ActiveCfg = Debug|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|Mixed Platforms.Build.0 = Debug|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|Win32.ActiveCfg = Debug|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|Win32.Build.0 = Debug|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|x64.ActiveCfg = Debug|x64
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Debug|x64.Build.0 = Debug|x64
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|Any CPU.ActiveCfg = Release|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|Mixed Platforms.ActiveCfg = Release|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|Mixed Platforms.Build.0 = Release|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|Win32.ActiveCfg = Release|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|Win32.Build.0 = Release|Win32
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|x64.ActiveCfg = Release|x64
+		{CF267D06-BE5D-4303-A013-33BBC076809F}.Release|x64.Build.0 = Release|x64
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Debug|Mixed Platforms.ActiveCfg = Debug|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Debug|Win32.ActiveCfg = Debug|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Debug|x64.ActiveCfg = Debug|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Release|Mixed Platforms.ActiveCfg = Release|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Release|Win32.ActiveCfg = Release|Any CPU
+		{1FC98FC2-84C9-4253-85B5-7BCC90D8F2E4}.Release|x64.ActiveCfg = Release|Any CPU
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
+	EndGlobalSection
+EndGlobal
diff --git a/windows/mxnet/mxnet.vcxproj b/windows/mxnet/mxnet.vcxproj
new file mode 100644
index 000000000000..def497554e9c
--- /dev/null
+++ b/windows/mxnet/mxnet.vcxproj
@@ -0,0 +1,337 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\include\mxnet\base.h" />
+    <ClInclude Include="..\..\include\mxnet\c_api.h" />
+    <ClInclude Include="..\..\include\mxnet\engine.h" />
+    <ClInclude Include="..\..\include\mxnet\io.h" />
+    <ClInclude Include="..\..\include\mxnet\kvstore.h" />
+    <ClInclude Include="..\..\include\mxnet\ndarray.h" />
+    <ClInclude Include="..\..\include\mxnet\operator.h" />
+    <ClInclude Include="..\..\include\mxnet\resource.h" />
+    <ClInclude Include="..\..\include\mxnet\storage.h" />
+    <ClInclude Include="..\..\include\mxnet\symbolic.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\base.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\cuda\reduce.cuh" />
+    <ClInclude Include="..\..\mshadow\mshadow\cuda\tensor_gpu-inl.cuh" />
+    <ClInclude Include="..\..\mshadow\mshadow\dot_engine-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\expression.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\expr_engine-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\expr_scalar-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\broadcast.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\channel_pool.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\channel_unpool.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\concat.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\crop.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\mirror.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\pack_col2patch.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\pad.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\reduceto1d.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\reshape.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\spatial_pool.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\spatial_unpool.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\swapaxis.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\extension\unpack_patch2col.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\io.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\logging.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\random.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\sse-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\stream_gpu-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\tensor.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_blob.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_container.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_cpu-inl.h" />
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_gpu-inl.h" />
+    <ClInclude Include="..\..\src\common\cuda_utils.h" />
+    <ClInclude Include="..\..\src\common\lazy_alloc_array.h" />
+    <ClInclude Include="..\..\src\common\object_pool.h" />
+    <ClInclude Include="..\..\src\common\utils.h" />
+    <ClInclude Include="..\..\src\engine\engine_impl.h" />
+    <ClInclude Include="..\..\src\engine\stream_manager.h" />
+    <ClInclude Include="..\..\src\engine\threaded_engine.h" />
+    <ClInclude Include="..\..\src\engine\thread_pool.h" />
+    <ClInclude Include="..\..\src\io\image_augmenter.h" />
+    <ClInclude Include="..\..\src\io\image_recordio.h" />
+    <ClInclude Include="..\..\src\io\inst_vector.h" />
+    <ClInclude Include="..\..\src\io\iter_batchloader.h" />
+    <ClInclude Include="..\..\src\io\iter_prefetcher.h" />
+    <ClInclude Include="..\..\src\kvstore\kvstore_local.h" />
+    <ClInclude Include="..\..\src\ndarray\ndarray_function-inl.h" />
+    <ClInclude Include="..\..\src\ndarray\ndarray_function.h" />
+    <ClInclude Include="..\..\src\operator\activation-inl.h" />
+    <ClInclude Include="..\..\src\operator\batch_norm-inl.h" />
+    <ClInclude Include="..\..\src\operator\concat-inl.h" />
+    <ClInclude Include="..\..\src\operator\convolution-inl.h" />
+    <ClInclude Include="..\..\src\operator\cudnn_activation-inl.h" />
+    <ClInclude Include="..\..\src\operator\cudnn_convolution-inl.h" />
+    <ClInclude Include="..\..\src\operator\cudnn_lrn-inl.h" />
+    <ClInclude Include="..\..\src\operator\cudnn_pooling-inl.h" />
+    <ClInclude Include="..\..\src\operator\dropout-inl.h" />
+    <ClInclude Include="..\..\src\operator\elementwise_binary_op-inl.h" />
+    <ClInclude Include="..\..\src\operator\elementwise_sum-inl.h" />
+    <ClInclude Include="..\..\src\operator\fully_connected-inl.h" />
+    <ClInclude Include="..\..\src\operator\leaky_relu-inl.h" />
+    <ClInclude Include="..\..\src\operator\lrn-inl.h" />
+    <ClInclude Include="..\..\src\operator\mshadow_op.h" />
+    <ClInclude Include="..\..\src\operator\operator_common.h" />
+    <ClInclude Include="..\..\src\operator\param.h" />
+    <ClInclude Include="..\..\src\operator\pooling-inl.h" />
+    <ClInclude Include="..\..\src\operator\reshape-inl.h" />
+    <ClInclude Include="..\..\src\operator\softmax-inl.h" />
+    <ClInclude Include="..\..\src\storage\cpu_device_storage.h" />
+    <ClInclude Include="..\..\src\storage\gpu_device_storage.h" />
+    <ClInclude Include="..\..\src\storage\naive_storage_manager.h" />
+    <ClInclude Include="..\..\src\storage\pinned_memory_storage.h" />
+    <ClInclude Include="..\..\src\storage\pooled_storage_manager.h" />
+    <ClInclude Include="..\..\src\storage\storage_manager.h" />
+    <ClInclude Include="..\..\src\symbol\graph_algorithm.h" />
+    <ClInclude Include="..\..\src\symbol\graph_executor.h" />
+    <ClInclude Include="..\..\src\symbol\graph_memory_allocator.h" />
+    <ClInclude Include="..\..\src\symbol\static_graph.h" />
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\src\c_api.cc" />
+    <ClCompile Include="..\..\src\engine\engine.cc" />
+    <ClCompile Include="..\..\src\engine\naive_engine.cc" />
+    <ClCompile Include="..\..\src\engine\threaded_engine.cc" />
+    <ClCompile Include="..\..\src\engine\threaded_engine_perdevice.cc" />
+    <ClCompile Include="..\..\src\engine\threaded_engine_pooled.cc" />
+    <ClCompile Include="..\..\src\io\io.cc" />
+    <ClCompile Include="..\..\src\io\iter_image_recordio.cc" />
+    <ClCompile Include="..\..\src\io\iter_mnist.cc" />
+    <ClCompile Include="..\..\src\kvstore\kvstore.cc" />
+    <ClCompile Include="..\..\src\ndarray\ndarray.cc" />
+    <ClCompile Include="..\..\src\ndarray\ndarray_function.cc" />
+    <ClCompile Include="..\..\src\operator\activation.cc" />
+    <ClCompile Include="..\..\src\operator\batch_norm.cc" />
+    <ClCompile Include="..\..\src\operator\concat.cc" />
+    <ClCompile Include="..\..\src\operator\convolution.cc" />
+    <ClCompile Include="..\..\src\operator\dropout.cc" />
+    <ClCompile Include="..\..\src\operator\elementwise_binary_op.cc" />
+    <ClCompile Include="..\..\src\operator\elementwise_sum.cc" />
+    <ClCompile Include="..\..\src\operator\fully_connected.cc" />
+    <ClCompile Include="..\..\src\operator\leaky_relu.cc" />
+    <ClCompile Include="..\..\src\operator\lrn.cc" />
+    <ClCompile Include="..\..\src\operator\operator.cc" />
+    <ClCompile Include="..\..\src\operator\pooling.cc" />
+    <ClCompile Include="..\..\src\operator\reshape.cc" />
+    <ClCompile Include="..\..\src\operator\softmax.cc" />
+    <ClCompile Include="..\..\src\resource.cc" />
+    <ClCompile Include="..\..\src\storage\storage.cc" />
+    <ClCompile Include="..\..\src\symbol\graph_executor.cc" />
+    <ClCompile Include="..\..\src\symbol\static_graph.cc" />
+    <ClCompile Include="..\..\src\symbol\symbol.cc" />
+  </ItemGroup>
+  <ItemGroup>
+    <CudaCompile Include="..\..\src\ndarray\ndarray_function.cu" />
+    <CudaCompile Include="..\..\src\operator\activation.cu" />
+    <CudaCompile Include="..\..\src\operator\batch_norm.cu" />
+    <CudaCompile Include="..\..\src\operator\concat.cu" />
+    <CudaCompile Include="..\..\src\operator\convolution.cu" />
+    <CudaCompile Include="..\..\src\operator\dropout.cu" />
+    <CudaCompile Include="..\..\src\operator\elementwise_binary_op.cu" />
+    <CudaCompile Include="..\..\src\operator\elementwise_sum.cu" />
+    <CudaCompile Include="..\..\src\operator\fully_connected.cu" />
+    <CudaCompile Include="..\..\src\operator\leaky_relu.cu" />
+    <CudaCompile Include="..\..\src\operator\lrn.cu" />
+    <CudaCompile Include="..\..\src\operator\pooling.cu" />
+    <CudaCompile Include="..\..\src\operator\reshape.cu" />
+    <CudaCompile Include="..\..\src\operator\softmax.cu" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\..\dmlc-core\windows\dmlc\dmlc.vcxproj">
+      <Project>{6e6dda36-69ba-4555-ba88-10166b11f2d2}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{15D3B5AC-7EF8-422E-BB67-69F06BA394C3}</ProjectGuid>
+    <RootNamespace>mxnet</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+    <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 7.0.props" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;WIN64;_DEBUG;_CONSOLE;MSHADOW_FORCE_STREAM;MXNET_USE_OPENCV;MSHADOW_USE_CUDNN;MXNET_USE_THREADED_ENGINE;MSHADOW_USE_CBLAS=1;MSHADOW_USE_MKL=0;USE_STATIC_MKL=0;DMLC_USE_CXX11;MSHADOW_IN_CXX11;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>../../windows/3rdparty/include;../../dmlc-core/include;../../mshadow;../../include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>libopenblas.dll.a;opencv_world300.lib;cudnn.lib;cudart.lib;curand.lib;cublas.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>../../windows/3rdparty/lib;%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
+    </Link>
+    <PostBuildEvent>
+      <Command>cd ../../python
+python setup.py install
+if not exist "$(SolutionDir)$(Platform)\$(Configuration)\opencv_world300.dll" (
+echo Copy dll files...
+copy ..\windows\3rdparty\bin\*.dll $(SolutionDir)$(Platform)\$(Configuration)\ /y
+)</Command>
+    </PostBuildEvent>
+    <CudaCompile>
+      <TargetMachinePlatform>64</TargetMachinePlatform>
+    </CudaCompile>
+    <PreBuildEvent>
+      <Command>@echo off&amp;SetLocal EnableDelayedExpansion
+cd ../../python/mxnet/
+(findstr /n .* libinfo.py) &gt; temp.txt
+(For /f "delims=: tokens=1*" %%a in (temp.txt) do (
+	echo.%%b| Find /i "        vs_configuration = 'Release'" &gt;nul
+	If !errorLevel!==0 (echo         vs_configuration = 'Debug') Else (echo.%%b)
+))&gt;libinfo.py
+del temp.txt</Command>
+    </PreBuildEvent>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;WIN64;NDEBUG;_CONSOLE;MSHADOW_FORCE_STREAM;MXNET_USE_OPENCV;MSHADOW_USE_CUDNN;MXNET_USE_THREADED_ENGINE;MSHADOW_USE_CBLAS=1;MSHADOW_USE_MKL=0;USE_STATIC_MKL=0;DMLC_USE_CXX11;MSHADOW_IN_CXX11;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>../../windows/3rdparty/include;../../dmlc-core/include;../../mshadow;../../include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>libopenblas.dll.a;opencv_world300.lib;cudnn.lib;cudart.lib;curand.lib;cublas.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>../../windows/3rdparty/lib;%(AdditionalLibraryDirectories)</AdditionalLibraryDirectories>
+    </Link>
+    <PostBuildEvent>
+      <Command>cd ../../python
+python setup.py install
+if not exist "$(SolutionDir)$(Platform)\$(Configuration)\opencv_world300.dll" (
+echo Copy dll files...
+copy ..\windows\3rdparty\bin\*.dll $(SolutionDir)$(Platform)\$(Configuration)\ /y
+)</Command>
+    </PostBuildEvent>
+    <CudaCompile>
+      <TargetMachinePlatform>64</TargetMachinePlatform>
+      <CodeGeneration>compute_30,sm_30;</CodeGeneration>
+      <FastMath>true</FastMath>
+      <AdditionalCompilerOptions>
+      </AdditionalCompilerOptions>
+    </CudaCompile>
+    <PreBuildEvent>
+      <Command>@echo off&amp;SetLocal EnableDelayedExpansion
+cd ../../python/mxnet/
+(findstr /n .* libinfo.py) &gt; temp.txt
+(For /f "delims=: tokens=1*" %%a in (temp.txt) do (
+	echo.%%b| Find /i "        vs_configuration = 'Debug'" &gt;nul
+	If !errorLevel!==0 (echo         vs_configuration = 'Release') Else (echo.%%b)
+))&gt;libinfo.py
+del temp.txt</Command>
+    </PreBuildEvent>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+    <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 7.0.targets" />
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/windows/mxnet/mxnet.vcxproj.filters b/windows/mxnet/mxnet.vcxproj.filters
new file mode 100644
index 000000000000..0c351fd1ef9f
--- /dev/null
+++ b/windows/mxnet/mxnet.vcxproj.filters
@@ -0,0 +1,442 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="mshadow">
+      <UniqueIdentifier>{a8ca4cfa-040b-447a-ac79-b7d1f9835624}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="mshadow\cuda">
+      <UniqueIdentifier>{d5e511b9-81d9-4890-8a51-262d4567ded8}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="mshadow\extension">
+      <UniqueIdentifier>{d7396bcb-df71-4492-b3a9-3ad1a99dc956}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="common">
+      <UniqueIdentifier>{0d84f9f7-bd5a-46b1-9ff1-3fd31483a8af}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="engine">
+      <UniqueIdentifier>{935c4f13-ffdc-4cde-bc1c-6ee9b424e951}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="io">
+      <UniqueIdentifier>{74864d56-248c-4191-8ada-c35ee9041de0}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="kvstore">
+      <UniqueIdentifier>{a91f9244-a58b-425e-90db-a7b71652e6f7}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="ndarray">
+      <UniqueIdentifier>{1abb0e4c-e4b7-4a64-b4b9-ba98c0887168}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="operator">
+      <UniqueIdentifier>{8129570b-2a10-4df0-90e6-78acdc72d147}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="storage">
+      <UniqueIdentifier>{5053e9c0-c5fa-4920-936b-6e4103b46fbb}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="symbol">
+      <UniqueIdentifier>{7d98986e-9f73-4010-b4e7-7f9518948b23}</UniqueIdentifier>
+    </Filter>
+    <Filter Include="include">
+      <UniqueIdentifier>{adba0dd3-a553-465d-a1f3-cf90e8efbd99}</UniqueIdentifier>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\mshadow\mshadow\tensor.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_blob.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_container.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_cpu-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\tensor_gpu-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\base.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\dot_engine-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\expr_engine-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\expr_scalar-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\expression.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\io.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\logging.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\random.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\sse-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\stream_gpu-inl.h">
+      <Filter>mshadow</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\cuda\tensor_gpu-inl.cuh">
+      <Filter>mshadow\cuda</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\cuda\reduce.cuh">
+      <Filter>mshadow\cuda</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\reduceto1d.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\reshape.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\spatial_pool.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\spatial_unpool.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\swapaxis.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\unpack_patch2col.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\broadcast.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\channel_pool.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\channel_unpool.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\concat.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\crop.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\mirror.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\pack_col2patch.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\mshadow\mshadow\extension\pad.h">
+      <Filter>mshadow\extension</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\common\utils.h">
+      <Filter>common</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\common\cuda_utils.h">
+      <Filter>common</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\common\lazy_alloc_array.h">
+      <Filter>common</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\common\object_pool.h">
+      <Filter>common</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\engine\threaded_engine.h">
+      <Filter>engine</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\engine\engine_impl.h">
+      <Filter>engine</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\engine\stream_manager.h">
+      <Filter>engine</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\engine\thread_pool.h">
+      <Filter>engine</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\io\iter_prefetcher.h">
+      <Filter>io</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\io\image_augmenter.h">
+      <Filter>io</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\io\image_recordio.h">
+      <Filter>io</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\io\inst_vector.h">
+      <Filter>io</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\io\iter_batchloader.h">
+      <Filter>io</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\kvstore\kvstore_local.h">
+      <Filter>kvstore</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\ndarray\ndarray_function.h">
+      <Filter>ndarray</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\ndarray\ndarray_function-inl.h">
+      <Filter>ndarray</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\concat-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\convolution-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\cudnn_activation-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\cudnn_convolution-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\cudnn_lrn-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\cudnn_pooling-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\dropout-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\elementwise_binary_op-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\elementwise_sum-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\fully_connected-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\leaky_relu-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\lrn-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\mshadow_op.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\operator_common.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\param.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\pooling-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\reshape-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\softmax-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\activation-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\operator\batch_norm-inl.h">
+      <Filter>operator</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\storage_manager.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\cpu_device_storage.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\gpu_device_storage.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\naive_storage_manager.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\pinned_memory_storage.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\storage\pooled_storage_manager.h">
+      <Filter>storage</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\symbol\graph_memory_allocator.h">
+      <Filter>symbol</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\symbol\static_graph.h">
+      <Filter>symbol</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\symbol\graph_algorithm.h">
+      <Filter>symbol</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\src\symbol\graph_executor.h">
+      <Filter>symbol</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\c_api.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\engine.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\io.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\kvstore.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\ndarray.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\operator.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\resource.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\storage.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\symbolic.h">
+      <Filter>include</Filter>
+    </ClInclude>
+    <ClInclude Include="..\..\include\mxnet\base.h">
+      <Filter>include</Filter>
+    </ClInclude>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\src\engine\threaded_engine.cc">
+      <Filter>engine</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\engine\threaded_engine_perdevice.cc">
+      <Filter>engine</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\engine\threaded_engine_pooled.cc">
+      <Filter>engine</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\engine\engine.cc">
+      <Filter>engine</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\engine\naive_engine.cc">
+      <Filter>engine</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\io\iter_image_recordio.cc">
+      <Filter>io</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\io\iter_mnist.cc">
+      <Filter>io</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\io\io.cc">
+      <Filter>io</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\kvstore\kvstore.cc">
+      <Filter>kvstore</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\ndarray\ndarray.cc">
+      <Filter>ndarray</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\ndarray\ndarray_function.cc">
+      <Filter>ndarray</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\convolution.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\dropout.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\elementwise_binary_op.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\elementwise_sum.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\fully_connected.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\leaky_relu.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\lrn.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\operator.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\pooling.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\reshape.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\softmax.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\activation.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\batch_norm.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\operator\concat.cc">
+      <Filter>operator</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\storage\storage.cc">
+      <Filter>storage</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\symbol\static_graph.cc">
+      <Filter>symbol</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\symbol\symbol.cc">
+      <Filter>symbol</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\symbol\graph_executor.cc">
+      <Filter>symbol</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\src\resource.cc" />
+    <ClCompile Include="..\..\src\c_api.cc" />
+  </ItemGroup>
+  <ItemGroup>
+    <CudaCompile Include="..\..\src\ndarray\ndarray_function.cu">
+      <Filter>ndarray</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\convolution.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\dropout.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\elementwise_binary_op.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\elementwise_sum.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\fully_connected.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\leaky_relu.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\lrn.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\pooling.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\reshape.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\softmax.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\activation.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\batch_norm.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+    <CudaCompile Include="..\..\src\operator\concat.cu">
+      <Filter>operator</Filter>
+    </CudaCompile>
+  </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/windows/mxnet/mxnet.vcxproj.user b/windows/mxnet/mxnet.vcxproj.user
new file mode 100644
index 000000000000..ef5ff2a1fae6
--- /dev/null
+++ b/windows/mxnet/mxnet.vcxproj.user
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="12.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <PropertyGroup />
+</Project>
\ No newline at end of file
diff --git a/windows/python/python.pyproj b/windows/python/python.pyproj
new file mode 100644
index 000000000000..06f35cc967e3
--- /dev/null
+++ b/windows/python/python.pyproj
@@ -0,0 +1,106 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003" DefaultTargets="Build">
+  <PropertyGroup>
+    <Configuration Condition=" '$(Configuration)' == '' ">Debug</Configuration>
+    <SchemaVersion>2.0</SchemaVersion>
+    <ProjectGuid>{1fc98fc2-84c9-4253-85b5-7bcc90d8f2e4}</ProjectGuid>
+    <ProjectHome>..\..\</ProjectHome>
+    <StartupFile>tests\python\test_tutorial.py</StartupFile>
+    <SearchPath />
+    <WorkingDirectory>.</WorkingDirectory>
+    <OutputPath>.</OutputPath>
+    <ProjectTypeGuids>{888888a0-9f3d-457c-b088-3a5042f75d52}</ProjectTypeGuids>
+    <LaunchProvider>Standard Python launcher</LaunchProvider>
+    <InterpreterId>{9a7a9026-48c1-4688-9d5d-e5699d47d074}</InterpreterId>
+    <InterpreterVersion>2.7</InterpreterVersion>
+    <EnableNativeCodeDebugging>True</EnableNativeCodeDebugging>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)' == 'Debug'" />
+  <PropertyGroup Condition="'$(Configuration)' == 'Release'" />
+  <PropertyGroup>
+    <VisualStudioVersion Condition=" '$(VisualStudioVersion)' == '' ">10.0</VisualStudioVersion>
+    <PtvsTargetsFile>$(MSBuildExtensionsPath32)\Microsoft\VisualStudio\v$(VisualStudioVersion)\Python Tools\Microsoft.PythonTools.targets</PtvsTargetsFile>
+  </PropertyGroup>
+  <ItemGroup>
+    <Compile Include="dmlc-core\doc\conf.py" />
+    <Compile Include="dmlc-core\doc\sphinx_util.py" />
+    <Compile Include="dmlc-core\scripts\lint.py" />
+    <Compile Include="dmlc-core\tracker\dmlc_local.py" />
+    <Compile Include="dmlc-core\tracker\dmlc_mpi.py" />
+    <Compile Include="dmlc-core\tracker\dmlc_sge.py" />
+    <Compile Include="dmlc-core\tracker\dmlc_yarn.py" />
+    <Compile Include="dmlc-core\tracker\tracker.py" />
+    <Compile Include="dmlc-core\yarn\run_hdfs_prog.py" />
+    <Compile Include="doc\conf.py" />
+    <Compile Include="doc\sphinx_util.py" />
+    <Compile Include="example\cifar10\cifar10.py" />
+    <Compile Include="example\imagenet\alexnet.py" />
+    <Compile Include="example\imagenet\data.py" />
+    <Compile Include="example\mnist\data.py" />
+    <Compile Include="example\mnist\lenet.py" />
+    <Compile Include="example\mnist\mlp.py" />
+    <Compile Include="example\python-howto\data_iter.py" />
+    <Compile Include="example\python-howto\multiple_outputs.py" />
+    <Compile Include="python\mxnet\base.py" />
+    <Compile Include="python\mxnet\callback.py" />
+    <Compile Include="python\mxnet\context.py" />
+    <Compile Include="python\mxnet\executor.py" />
+    <Compile Include="python\mxnet\initializer.py" />
+    <Compile Include="python\mxnet\io.py" />
+    <Compile Include="python\mxnet\kvstore.py" />
+    <Compile Include="python\mxnet\libinfo.py" />
+    <Compile Include="python\mxnet\metric.py" />
+    <Compile Include="python\mxnet\misc.py" />
+    <Compile Include="python\mxnet\model.py" />
+    <Compile Include="python\mxnet\name.py" />
+    <Compile Include="python\mxnet\ndarray.py" />
+    <Compile Include="python\mxnet\optimizer.py" />
+    <Compile Include="python\mxnet\random.py" />
+    <Compile Include="python\mxnet\symbol.py" />
+    <Compile Include="python\mxnet\visualization.py" />
+    <Compile Include="python\mxnet\__init__.py" />
+    <Compile Include="python\setup.py" />
+    <Compile Include="tests\python\common\get_data.py" />
+    <Compile Include="tests\python\common\models.py" />
+    <Compile Include="tests\python\test_tutorial.py">
+      <SubType>Code</SubType>
+    </Compile>
+    <Compile Include="tests\python\train\common.py" />
+    <Compile Include="tests\python\train\test_conv.py" />
+    <Compile Include="tests\python\train\test_mlp.py" />
+    <Compile Include="tests\python\unittest\common.py" />
+    <Compile Include="tests\python\unittest\test_bind.py" />
+    <Compile Include="tests\python\unittest\test_infer_shape.py" />
+    <Compile Include="tests\python\unittest\test_io.py" />
+    <Compile Include="tests\python\unittest\test_kvstore.py" />
+    <Compile Include="tests\python\unittest\test_ndarray.py" />
+    <Compile Include="tests\python\unittest\test_operator.py" />
+    <Compile Include="tests\python\unittest\test_random.py" />
+    <Compile Include="tests\python\unittest\test_symbol.py" />
+  </ItemGroup>
+  <ItemGroup>
+    <Folder Include="dmlc-core\" />
+    <Folder Include="dmlc-core\doc" />
+    <Folder Include="dmlc-core\scripts" />
+    <Folder Include="dmlc-core\tracker" />
+    <Folder Include="dmlc-core\yarn" />
+    <Folder Include="doc" />
+    <Folder Include="example\" />
+    <Folder Include="example\cifar10" />
+    <Folder Include="example\imagenet" />
+    <Folder Include="example\mnist" />
+    <Folder Include="example\python-howto" />
+    <Folder Include="python" />
+    <Folder Include="python\mxnet" />
+    <Folder Include="tests\" />
+    <Folder Include="tests\python\" />
+    <Folder Include="tests\python\common" />
+    <Folder Include="tests\python\train" />
+    <Folder Include="tests\python\unittest" />
+  </ItemGroup>
+  <ItemGroup>
+    <InterpreterReference Include="{9a7a9026-48c1-4688-9d5d-e5699d47d074}\2.7" />
+  </ItemGroup>
+  <Import Project="$(PtvsTargetsFile)" Condition="Exists($(PtvsTargetsFile))" />
+  <Import Project="$(MSBuildToolsPath)\Microsoft.Common.targets" Condition="!Exists($(PtvsTargetsFile))" />
+</Project>
\ No newline at end of file