most of this is recycled bits...
There are several major hurdles to face. Keep in mind that most of the online articles I've read on the RPD state the belt feed design is a rather weak one that was prone to misfeeds. SO if those articles are to be believed, you've got a tenous design to begin with!
Unmodified M60 links won't work due to the taper of the 7.62x39 cartridge. Notice I said "unmodified"... you could probably stretch the bottom half of the links with a linking "tool" designed for the job. The links, once bent out, would not fit .308 and would need to be hand-fit... every single one of them... to make sure they fit and feed the 7.62x39. Conclusion: Buy cheap & available RPD belts
Stripping the bullet from the belt isn't a problem because the RPD is a "push" type, like an M60, like a normal AK, and therefore a normal AK bolt will work, assuming you locate the belt at the same location just below the ejector like a normal magazine would feed. That's the easy part... your starting point is the bolt and feed mechanism.
Once the belt is located properly, this should cause the major changes to become obvious. First, you happen to have a receiver shell on both sides of the belt. You've got to cut a hole in both sides to get the belt to get it to fit. Better yet, build a new receiver that has more flexibility.
You'll need a feed tray to ensure the belt reliably feeds into position without kinking/jamming before the feed pawl grabs it. Are you going to make this able to double-feed from a magazine as well as a belt? How are you going to let the magazine clear the feed tray? Swing-away feed tray? Trap door?
Next, you've got to figure out how to create a feed pawl that pulls the belt into position, locks it in place while the bolt cycles, and feeds once the round is ejected. Attaching a cam actuator (lifter lug?) to the bolt carrier is really the only way to do this. But where? Obviously, on the side or on top. Look at those experimental belt-fed RPK museum pics posted last year to get what I mean.
Then you've got a complicated feed lever that has a complex curve which is motivated by the cam actuator on the bolt carrier. That means a smooth curve, not a 30-degree straight cut. Add to that the timing of exactly how wide that arm needs to be. That's the only part that I see - you must craft a nice smooth curve into either a brake to bend a piece of channel steel, or machine the nice smooth curve into the metal. Maybe this can be simplified. Better yet, maybe an M60 arm could be used. For that matter, maybe an MG42/MG3 arm could be used... would be cheaper at least.
Put a housing on top of the whole thing to keep it from getting dirt in the mechanism.
Add a new top cover (minor issue).
NOW what you've got is an un-tuned belt-fed AK. I would strongly suspect you would need more impulse power from the gas port to ensure the extra mechanical mass can move. Yes, an AK is overpowered but how much energy will be needed to compel the feed mechanism to move? Inertia is NOT your friend! So maybe a larger gas port. Maybe an AKSU-style muzzle booster. Here's a thought - the normal upper hanguard has those distinctive grooves pressed into them. The Yugo M92 is the exception in that it uses an un-dimpled tube, and that has a piston "ring" (more like a collar" with holes in it.
Well, let's assume you'd want *all* of the power to cycle the action. Cover the piston bleed holes on the gas block (like some of the AK-74 style gas blocks). Use an upper handguard WITHOUT holes and all power would go to the action and none would bleed off. The only bleed would be out the front of the barrel. If the piston actually had a smooth tube to ride in, you could use piston rings like a car engine. Rings are used on the Steyr AUG's system, as well as the HK G36 gas system. That gives a total seal that lets the short-stroke piston work so well, and keeps the gun clean. In this case, the extra power (how much I wonder?) would let us work the feed mechanism without a hitch.
Stock AK barrel? No quick-change design means you need to let it cool between belts. An excess of 100-rounds rapid fire will NOT be kind to your barrel. Something to consider.
I didn't even mention the possibility that the extra belt-feed junk may require the FCG to be relocated. Maybe.
Can it be done? I'm certain of it - look again at those museum pics. Not sure how well it will work.... all due to the inherent power of the 7.62x39. Now, maybe all those reports I read were inaccurate. Then again, how many people bitch and moan about a G2 trigger hump causing issues? Will the feed mechanism impart more, or less, resistance than a trigger hump? You'll have to get that thing PRETTY SMOOTH! Frankly, a short-stroke system like an SKS (or G36 or AR-180) may be even better. The MG42 uses barrel recoil to begin the cycle of the action. That is something to consider - albeit a major redesign - float the barrel at the point when the bolt lugs rotate into their unlock position. A brand-new carrier would make that possible. The trunion would need to accomodate the headspace in such a way that, once headspaced, the barrel can be removed. Now, the PKM uses a little cover plate that is indexed to a notch in the barrel, so that all PKM barrels are all headspaced due to the interaction of the locking "lug" on the index plate. That idea could be adopted to the bolt. This would look less like an AK and more like a brand-new rifle. May not be necessary, and that totally depends upon testing unless you can figure how many foot-pounds of energy it would take to cycle the action with a belt, and compare that to the recoil energy. I know that's the proper way, and the "let's try and see" method shouldn't really be attempted.
It would be very interesting to see it happen no matter what. If you could get it 80% there, then you'd be set.