In 1975 I assembled a MITS Altair kit. I still have the machine, though I have not powered it up in many years. Discrete wiring of the front panel was memorably tedious and error-prone.
The machine had 256 bytes of static RAM on a board separate from the CPU board.
MITS later sold a 4K DRAM board, which if I recall correctly, tried to rely upon RC timed one-shots for its timing, and an incredibly poor PCB layout with numerous jumper wires. This board was a total botch. MITS later came out with a different board designed with synchronous logic that might have been better. Eventually MITS and others came out with larger static RAM boards that worked reliably.
My coolest project with that machine was to use an A/D / D/A board and some code hand-assembled to binary (and keyed in through the front panel switches), to make an audio delay line, whose delay time was controlled by the front panel switches. Running a radio straight into one side of stereo headphones, and the delayed audio into the other side, made for an amusing experience.