Steel And Concrete Don't Keep Well
The thing about steel is that no matter what you do it will eventually oxidize completely. In many ancient steel artifacts if you hold a magnet up to them nothing will happen. This is because iron-oxide has totally replaced the entirety of the artifact. Steel as a load bearing item within a shelter would eventually oxidize enough that it no longer provided structural support and it would buckle, then collapse. Likewise, concrete is porous and tends to absorb water over time. Freeze/thaw cycles make cracks form allowing more water in, and eventually it crumbles as well. In terms of archaeology.
In addition, a vacuum would not keep for more than maybe a few years, if that. As Willik stated, nature abhors a vacuum and keeping an atmosphere reduced to a vacuum is something that requires constant maintenance and effort to sustain. Eventually a seal somewhere would degrade enough to give way and the atmosphere would gradually equalize with the outside.
Keeping a Facilitiy operational requires Constant Maintenance
Anyone who has ever been in the military can recall endless hour upon boring hours simply cleaning, painting, and maintaining their surroundings. This is more than just busy work for the brass to keep the enlisted too busy to get into trouble. Large facilities require constant maintenance to remain useful and functioning. Once that maintenance stops it is usually only a matter of years until the site is useless for whatever purpose it was designed for. Nuclear missile silos that were only decommissioned a few decades ago look like this now:
Keep in mind that these are some of the most robust and high tech structures of this era. For ideas of what bunkers look like after just a century here is a portion of the maginot line from WW1:
Alternate method of preservation
So lets say this shelter was built high in the mountains close to a glacier. The nuclear war kicks off. A nearby blast broke one of the seals on the air filtration system and it cannot be repaired in time to prevent everyone inside from dying to fallout exposure. In a moment of delirium the last survivor throws open the main doors and collapses in the entrance and dies. In the next few months water fills the bunker and as nuclear winter kicks into full swing it freezes solid. Layer after layer of snow piles up on top of the bunker and eventually glaciation has buried the entire thing in snow and ice. The bunker lays that way, frozen in ice for a few thousand years. Obviously it is not in a perfect state of operational repair, that would not be possible. But it is still in a remarkable state of preservation. Eventually it is unearthed and becomes a major cultural site and archaeological study since no other place had such a high degree of preservation associated with it. Here is an image of a plane frozen in a glacier from WW2:
It was so well preserved they are actually repairing it to full working condition after almost a century under the ice. Perhaps it would be possible under the right conditions to freeze your bunker solid and leave it in a state of well enough preservation that your future archaeologists could reconstruct or repair parts of the bunker to re-produce functionality. It would be a very long stretch honestly, just finding a bunker well enough preserved after a thousand years to walk around inside and maybe find some intact skeletons would itself be an incredible occurrence. I do not see any way a bunker could be preserved in an operational state across such time-spans, but that is not necessary to make a site an incredible archaeological site. Even just finding a single intact body frozen in the ice like Otzi the iceman would be an astounding find. Since I am tossing so many pictures at you, here is Otzi the Iceman, incredibly preserved in ice along with all his personal effects and tools scattered around him when he died after 5300 years. The body was so well preserved that initially the police thought he was a murder victim or lost hiker from only a few years prior until they started finding copper age tools.