It is doable, easily, if you consider time-dependent EM fields - or, more specifically, pulses of light. This is because light can be focused into a tight focal spot of ~ wavelength size, thus squeezing a lot of energy (and thus EM field) into a tight spot. This is impossible using static fields, which can't reach maxima except at their sources. Just as importantly, one can prepare really short pulses of radiation that compress into a short (i.e. few femtoseconds) time the energy stored in a laser medium over a millisecond or so. This allows quite modest energies - of millijoule order - to create electric fields comparable with the atomic electric fields (as in John Rennie's answer) for long enough to ionize just about any atom.
Note that this does not need a macroscopic sample (although actually observing stuff usually does!). If you manage by some means to get a single atom to sit still at a known position (which is a doable but nontrivial exercise) then firing enough strong laser pulses will ionize it, releasing one electron. You can of course release more electrons but it will naturally get harder and harder (so that triple ionization is the most you can hope for). Photon energies (which are proportional to the frequency) are not really a concern in this high-intensity regime, where multiphoton ionization and tunnel ionization take over as the main mechanisms as one goes from optical to IR and lower frequencies.
Regarding the converse in your question, we have nothing like the technology to "atomically print" a table, atom by atom, and I would question the feasibility of such a scheme. Optical tweezers, for example, can be used to move molecules around and depositing them on a surface could potentially work, but their spatial resolution is limited (by the wavelength, which can't be too short or stuff starts to break). Even worse, you'd have to work at it for something like geological-scale times to get anything sizable.
On the other hand, metamaterials are a good example where we can build materials to atomic-scale detail so that they have the macroscopic properties we want, including such exotica as negative refraction indices or strongly chiral media. While it does not exactly fit the bill you ask for, I think it is fundamentally more cool with the added bonus of being feasible.