All the shapes and geometry were generated using the internal 3D shape primitives in combination with interesting sets of cloner systems. For example, triangles can be easily made by creating a cylinder with only 3 subdivisions along the circle face. Similarly, the hexagons are cylinders with only 6 subdivisions along the circle face.

In some cases, I used cloners within cloners to easily multiply the complexity by an exponential degree. This also saved a great amount of effort in iteration and animation due to the procedural nature of the system. By combining different types of effectors, the cloner systems could be easily varied and animated with color, size, position, and number of objects. 

Math Modifiers were used to create Perlin noise functions to drive some slow drift and create depth from the flat layout. Continuous modifiers were used to animate slow spinning of various elements around the circle. Delay modifiers driving Randomize Effectors and Turbulence Effectors created the staggering of animation for the intro / outro using a single set of keyframes. The slowly increasing flickering rate was controlled by a Math Modifier driving Colour Ramp effectors. 

In essence, the whole system was designed for easy adjustments for creating iterations, and intended to be animated by as few keyframes as possible.

Camera switching and cutting without keyframing a child selector or multiplexor involves leveraging the timeline in unison with vertical node hierarchy in the nodegraph. This is the closest I could get to a workflow similar to what I'm used to in After Effects for easy camera rearranging or re-timing. Varying the zoom / field of view parameter between cameras was also essential in creating a wider gamut of perspectives. I prefer to use cameras in what I call a "Null Sandwich", which allows the camera to easily orbit your parent Null while targeting a child Null, giving independent control of these parameters.

The motion graphics principles explored mainly involve easing and sequencing to result in fast movement across camera transitions. Something important I learned is that having the movement be continuous across a camera switch is essential to tie the sequence together instead of resulting in a sudden, jarring cut.

To create interest within 1 camera shot, the movement needs a slowdown or pause to allow the viewer to take in what you are presenting. 

As a result of these 2 principles in play, I ended up with keyframe easing that looks like S shapes, where the start and end of any camera shot has significantly more movement than the middle of the camera shot. The amount of movement at the start and end should also be similar in magnitude to the preceding or succeeding camera shots to allow for sensible transitions.

A big thanks to the resources of https://freesound.org/ and https://motionsound.io/ for the samples used here.

All sound design was done post-render in Adobe Audition.