I was the one who posted the thread a while back about this, and yes, concluded that it "just barely" works. I used this 100W UV LED surface-mount chip:
https://www.amazon.com/Chanzon-Ultraviolet-30V-34V-Components-Lighting/dp/B01DBZI4SA -- after trying a few different wavelengths, the 395nm seemed to work best as anything narrower seemed to get eaten up by the condensers and enlarging lens (I primarily used a f/3.5 Rodenstock El-Omegar but I understand the El-Nikkors also pass a decent amount of near-UV).
Focusing was not a major issue as I simply focused with the regular white light bulb before switching to the UV head which I designed and 3D-printed. My cyanotypes at least have all been decently sharp.
The bigger problem of course was heat causing the negative to warp in the glassless carrier. I solved that to some extent by making a double-sided glass carrier, but with 30-40 minute exposure times for an 8" square print, it still made me nervous about using any especially valuable negatives.
There were also some issues with light falloff at the corners which I suspect could be solved by tuning the distance between the LED and the condenser.
I put the project on hold for the time being because I managed to burn out my last LED chip by forgetting to turn on the cooling fan during an exposure... not a mistake one makes twice. However I think I've maximized the results from my Frankenstein'ed Omega B600 and need to move on to designing a custom negative stage, which would allow switching from condensers to a diffuser (Edmund Optics carries ground fused silica panels that should work splendidly), as well as adding a small cooling fan between the diffuser and the negative carrier.
Longer term I think the limiting factor will be the attenuation of the enlarging lens... I'm trying to study up on optical design to see how practical it would be to create a UV enlarging lens from some of the elements Edmunds carries that are color corrected down to 220nm.