The first attempt at designing a swerve drive, so feel free to let me know of anything I messed up… there are probably a few things.
The gearing is
14 to 24
22 to 22 (the 1:1 gearing was the keep the motor in the center)
15 to 60
A gear ratio of 6.86:1
Speed of roughly 12.39
The Versa is a 36:1 ratio with a 24 to 80 tooth pulley. The final ratio is 120:1 on the neo 550, turning at 91.6rpm
Nice first module. Lot to learn from the process 
There are a couple small things that don’t really matter (fall under the category of “I would do it such a way” or “community wisdom is ‘such and such’”) so I won’t dwell on any of that unless you want a stupid amount of minute and half useless feedback.
I am tracing the loads normal to the floor through the wheel, axle, and forks to the inner race of the bearing. I can see the bearing is retained from the bottom with 8 (I am guessing 8-32s). Where is this loading transferring after this (how does the bearing transfer the loads to the bottom plate)? i.e. what prevents all the whole wheel, fork, and lower gearing from being shoved up into the top plate? I can’t tell with the provided screenshots.
Any advice also on making it lighter? Using 1/8 plate? or other options?
I haven’t really seen much swerve up close in a while, and this is the first time even researching it so I expect many things to be off.
You could use the UltraPlanetary gearbox with the Neo550 you have to save some weight and size.
I would also add a web of material on your main plate between the area of the VP and pulley and the main module.
Re: Skye’s point above: Most teams deal with this by either having screw heads also on the top side, similar to the WCP module, or by having a counterbore in the plate that the bearing sits in so that the load transfers from the bearing into the main module plate. This requires a bit more machining capability, but if you have a router to cut your plates already, as long as you have decent z-axis control you should be fine to do this. We use an Omio router to do ours and have not had issues with this method.
Overall, I and many others could probably give more advice if you have a CAD model you could give to allow us to dive a little deeper into your design. I also suggest searching around here on CD to look at what some other teams have done, and how swerve design has evolved over time. Looking at a lot of posts that Aren Hill, Bomb Squad, Kevin from 2451, and others will show a lot on how designs have been iterated over time to be improved.
For current methods that represent many of the optimized modules out there today, the Swerve Drive Specialties and West Coast Products modules are both good COTS examples, while modules from 1690, 2471, 2767, and (shameless plug) my own are some good examples of custom designed modules.
Swerve is fun and its one of the more interesting design products that can help give you a project to sharpen your CAD and design skills, while being something that is productive and be actually usable down the road.
Onshape Link
It may be a bit brought in some parts of the part studios, there was a lot of backtracking, confusion, subtle cursing, and such.
As for the extraplanetary I haven’t played with them or really looked into them so it slipped my mind.
Make sure you have a way to sense the absolute position of the module. Since you have a reduction between the VP encoder and the module, the VP encoder won’t tell you the true position. You would need a zero sensor, start the module in a consistent place, or add an absolute encoder.
Yeah, you need a plate or some shoulder feature for the axial load on that bearing to push up against in order to transfer the loads to the frame without putting stress on the powertrain shafts.
You can use your existing bearing retention screws on the bottom for this task (just make a clearance hole in your larger (existing) plate.
Here is what it would probably look like in cross section (blue is on the crossectional plane, black is behind it) the end result is a part that looks like a ring with a bolt circle in it. (Edit - The mark-up I did here is slightly off, but the concept is fine, just go to the next image)
…and a (bad) orthographic view, you could easily tie into the red standoffs (dashed line)
