The unstable point is at the top of the crest. You would need a vacuum chamber with dampened vibrations to have any chance of it standing there, as any small deviation from a perfectly balanced position would make it fall.
In practice, friction and material deformation will probably give the top of the crest some degree of stability, possibly enough for a demonstration. At the very least it will probably be much slower to roll from such a metastable orientation.
Edit: also, based on the following gif someone else posted, I don't think the top of the crest is the metastable point. You can see the object linger at that point before seemingly spontaneously finding true equilibrium - illustrating my point rather well.