Check the extractor to see if it holds the round tight, try just removing it for cleaning & inspecting the parts for wear when apart.
And check the ejector to see how deep it sets in the groove in the side of the bolt it may not be deep enough. It needs to almost be touching the bottom of the groove in the side of the bolt.

got one in the shop with same problem. This is the one I'd previously inquired on gas port diameter, thinking it might be under-gassed, despite being a factory barrel. Was 3.5mm I'm still messing with it, but it appears the ejector may be too short. The one I have is a NODAK which makes a dimensional issue less likely than yours, but it is still all I have left to look at.

take off the cover and the spring then cycle it to where the ejector is in the slot of the bolt and carrier and see how much clearance you have on the bolt you should not have much at all if its more than the thickness of a hair (roughly) then you might have to much... Take a pic of that and post it Im sure we all will give you our views... you can check it with a Caliper and see what the measurement is from the outside of the receiver to the outer most part of the extractor. If you need that dimension let me know for a Ak-47..

It is difficult to tell from the pic but I see too issues - one the ejector needs to be longer. The cause of this could be the ejector itself, or the center support rivet could be too long which draws the ejector away from center-line. The bolt should be as close to the ejector as possible. It is a weld repair. I weld them so they are too long and file until they just kiss the bolt.

The other issue, again difficult to be sure from the pic, is the ejector appears to be curved on its leading edge. It should be as close to 90 degrees as possible. I'm speculating that the slight concavity is contributing to an ejection path more forward than to the right. Perhaps squaring this wil fix the problem without any welding.

If you do weld, I use a copper backing plate and stainless weld. The lower carbon content in the stainless, at least theoretically, will reduce the chance of carbon migrating to the heat and making the weld brittle.

If one does a poor job of the welding, then yes. If one does a good job then it will be just fine. This is a factory authorized repair, and while I don't know how to translate the European specifications for the weld material from the original Russian and East German manuals, I've had no issue with using ER 308/308L at 75 amps (TIG).

piston is supposed to wiggle - a design feature, not sloppy machining.

The piston + carrier overall length is to be in a specific range. If it is not, such as with people using a '74 piston on a 47, one may have problems. But changing the piston length without knowing what it is supposed to be makes no sense to me.

Maybe these will help you. They show the proper bolt/ejector clearance.

I don't recall who originally posted that spec drawing.....it's been awhile. It may have been RForbus. He does some really good CAD work.

yeah, the 34mm really doesn't appear to be right for a 1mm receiver, but that's only off by about .004" per side.

If measurements were taken off a receiver made from slightly off-spec material, that may account for that discrepancy. I really don't know.

QUE?