It's very different between the Khronos standards (including Vulkan) and DirectX.
In DirectX, Microsoft implements the API, but they publish to GPU vendors a HAL API. There's actually two HALs: one that runs in kernel-mode, to communicate with the card directly; and one that runs in user-space, to do other tasks (like manage memory, set up data structures, and compile shaders).
In Vulkan, Khronos only publishes the specification and a test suite. It's up to each GPU vendor to ship an implementation of the library. The application's interface to the driver is a normal C API, like any other library. You'll get confused if you try to think of an "API part" communicating with a "graphics card driver": the driver is the API implementation.
This is why it's very hard for third-parties to make open-source OpenGL (or Vulkan) implementations: you need to know all the details of the GPU hardware to do it. These graphics APIs are defined by industry groups consisting of the same people who design the hardware (at a high level), and deep understanding of the GPU architecture is what you need to design an efficient graphics API.