I assembled the keypad and put on a range of keycaps (mostly test prints, but also some off-the-shelf keycaps my roommate had lying around).

I tested it and it worked, but sometimes key presses were not registered. First I thought it was the switches not being closed, but it turned out to be a firmware problem. I used the ergodox_ez qmk firmware for the keypad, which is configured for a split keyboard. Normally the right side is connected via USB (as it is for the keypad), but it is also connected to the left side via a 4-pole 3.5mm jack, speaking I²C. This normally works great, however, as my keypad has no left side talking back, it waited for a response, which lead to a lower sampling rate and thus key presses not being registered.
After a lot of procrastination, I finally created a [custom firmware in qmk](https://github.com/AberDerBart/qmk_firmware) which was no big deal after all and now it works like a charm.

Next step: print and paint the final keycaps