Hello Janyk!
You've pointed out a good reason why your end result ends up looking off: you're not getting the foundation right, so everything else you build on top of it will look out of place in some way.
You could probably try doing facial gestures of 30-60 seconds each. Focus on the center line of the face (the vertical line that goes from the middle of the forehead down the center of the nose to the bottom center of the chin) and the horizontal lines (the height of eyes, the tip of the nose and the corners of the mouth).
Just a quick example:
https://i.imgur.com/tyVLxKv.png
You can also build on this: after finding the location of the eyes, nose and mouth, start adding the ears. The more time you have to spend on a gesture, the more detailed you can get. Focus on the jawline and hairline/head shape, and build from there. It will take some time to get comfortable with this, especially because the point is not to draw refined, accurate looking faces. You want to get comfortable with the proportions and making quick decisions helps to train your brain for this.
If you have this as a correct foundation, you'll be able to draw faces that look just right before you even know it. :)