>as far as we know...
This is exactly why I am impressed. I have some basic understanding of rocket engine and aerospike design, and can recognize most of the features and the functions they would serve in this design. I have no clue how optimal they are, but it is clear that they are taking function into consideration.
Moving from the outside of the engine in, you can see that support lattice has been added to the exterior and density adjusted on the exterior to reenforce high stress areas.
Next, the coolant jacket around the combustion chambers has has baffles (verticals s curves) that seem to adjust cooling in specific areas.
It looks like the spike cooling system (the dense horizontal lines) is trying to balance fluid resistance with how much the cryogenic oxygen.
The one part I don't understand is what the Y shaped channels in the very center are supposed to do. This is the liquid oxygen flow to the tip of the spike before it returns along the outside of the spike as a heat exchanger (mentioned above). It looks like some sort of expansion chamber but I would have to ask a professional aerospike engineer what it is for.
This is a long way of saying the structures show functional intent to converge on some solution. Some parts seems random, but most algorithmic design is random change and iterative testing; Essentially evolutionary code, where permutations are made, compared, killed off, then fed into future models. I wonder how they handle so many degrees of freedom, and how much of what we see in the output was pre-defined.