I don't think it is that impressive anymore. We've been building rockets for decades. Making them return to Earth is peanuts compared to building a self driving car. You can even make a simulation that is 99% accurate without much effort. Also, rocket science is just Newtonian physics.
Of course, building a rocket requires a shit-ton of resources, so if anything is impressive then it's the management of those resources.
The newtonian physics part of flying a rocket is indeed the boring part of rocket science in these days of Ghz computing.
But all the engineering (an altogether different - if related - discipline) required is anything but simple.
And engineering and all of its sub-disciplines (materials science, propellant research, iterative refinement, operational research, logistics, 3d printing, computing, simulation, structural engineering, etc...) is both where the complexity lives and where the greatest progress in rocket science has been made.
Software engineers have a lack of self-esteem when comparing to other STEM disciplines. The reason we see more fuck-ups in software than in other fields is not because software engineers are stupid, but because software is inherently difficult.
Software engineers are not stupid, but in other STEM disciplines they have a reputation of making themselves look stupid because of beliefs like "it's just newtonian physics".
Of course, building a rocket requires a shit-ton of resources, so if anything is impressive then it's the management of those resources.