I presume that you're not attempting to motivate using such representations, you're simply presenting this as another formalism for handling standard phonological processes, and the point is to show how it works. It depends on how much you are willing to ignore or arbitrarily stipulate. I would advise against vowel harmonies as a first example, because they add the complexities of locality, underspecification, and no-crossing, in crossing consonants. Instead, I would start with strictly adjacent consonant-to-consonant assimilations like voicing and nasal place. However, this also depends on how much else you're willing to add to autosegmental theory, specifically whether you can get away with introducing the notion of feature grouping ("geometry") so as to have a "place" thing that can spread.
If that (grouping) would be too much, then I suggest sticking to single-feature segment-to-segment assimilations. Which would be voicing, basically. You might be able to include nasalization if you've already dealt with incidental feature changes (where voiced oral obstruents like /b/ become voiced nasal sonorants like [m] -- the obstruent change is incidental to nasal spreading). But I think you should simply deal with the question of grouping, since it is bound to come up in the question period and someone asks about spreading two features simultaneously. In which case, you will be set up to talk about nasal place assimilation, and you can have a much more interesting discussion of the predictions of linear vs. non-linear assimilation.