I think I should amend my post. When you work on your own art, don’t worry too much about artistic quality. Artistic quality is about getting shapes, volumes, shading right - making sure the asset “looks” good. This doesn’t mean you shouldn’t try to give your best, but the sort of quality that is often ignored by regular artists is technical quality, or the craftsmanship. i.e. that an asset doesn’t just look good, but that it is, indeed, overall a good and solid asset.
Smaller assets let you focus more on this aspect of the work.
Many tools you’ll be developing will support artists in achieving higher quality of craftsmanship. e.g. better edgeloops, better normal maps, less intrusive tools and more intuitive use of tools (UX comes into play here - having working on assets yourself will help with this a lot!).
About portfolio: I don’t think there’s ONE approach that works for every TA. It really depends on your specialization. And personally, while nice art impresses me, a TA folio full of art will not get you hired. I consider having some art pieces in the folio a bonus, but even there I look more for craftsmanship (if you’re nice enough to supply maps and wireframes) and for “what” you did. e.g. a concept drawing is less useful for me than a rigged character, or a PBR textured asset, or if you’re able to take your work into a game engine and present it there.
Since you’re a SW engineer, one thing I value a LOT is if people can apply good software engineering principles at their work. This is something difficult for many TAs because they either don’t have the training or awareness to do so (maybe because they come from an art background). Yet with tools and pipelines becoming more complex, it really pays off to have team members with such knowledge. Rob Galanakis is a good example of a TA who cares about this and who makes strong points for the needs of those skills in his blog. If you make a folio, you should definitely try to somehow convey these skills!