The most challenging part of the build is designing a custom chassis to fit proprietary NUC motherboards. The right tools help — CAD software gets you close — but there is no AI that can do this part for you. Not perfectly. Not the way you want it. This is millimeter-level work. Measure, design, print, test, adjust, and repeat until it fits exactly right.

The chassis is currently on version 2.2. Every successful print increments the version by 0.1 — each number represents a real iteration, a real adjustment, a real improvement. Weeks of work to get here. The chassis is finally mounted and fitted correctly, and now the focus shifts to upgrading the hardware inside it. 

This build pushes the boundaries of filament density and durability using economical generic PLA. PLA is sensitive to heat — it can warp or collapse under sustained thermal load — so the entire rack design is built around a semi-open enclosure that prioritizes airflow above everything else. If every component stays below 150 degrees, there is no failure. That is the theory and so far it holds.

Almost every part in this rack came from decommissioned legacy hardware. Repurposed and recommissioned devices are where most of the sourcing happens — brand new parts only get ordered when there is no other option. Each 1U slot in the rack is either a fan or a device. The current build plan targets 8 CPU nodes operating as a cluster for heuristic RF and network monitoring, feeding directly into the Kismet Project and AEGIS.