The 3d code you write in 3dgs isn't really much different than doing it in your own engine, just different function calls.
No, there is much of a difference. Just compare code written for 3dgs with code written for Irrlicht. Then you get a little hint. But if you compare to code written for Ogre the difference raises and last but not least the very well structured and object oriented code of the C4 engine is much of a difference.
The code looks completely different if you are using classes, templates, virtual objects, interpreted type conversions, singletons or other stuff that you will not find in a 3dgs code.
And there is a difference in C++, C# and Lite-C. The first 2 are supporting classes, inheritance and other modern techniques but you still have to care about pointers, references, memory leaks in C++.
You have to be a good software architect to create good, robust and reliable C++ code. It will be more easy to create good C# code even for a beginner.
And if you really want to create your own appealing demo then you might impress some friends but not a real game dev company. Such companies know about the real problems. Rendering and lighting a cube with a few OpenGL or DirectX commands are not really a problem. But collision-detection, scene-management, path-finding, dynamic shader creation on the fly, efficient shadow mapping, big terrains - these might be some impressive features.
If you come to an established company then they already use a technology. They probably need help to improve their scene-management, their shader library, their AI. They might be interested in converting a title to a game console. That is what you could offer them.
