I re-carved it with the bit that cuts better. I can't drill holes yet because I have to carve the other side first.

Here's the result, next to the previous result for comparison.

Before starting the other side of that, I decided I'd better make some test cuts, and good thing I did as I discovered something interesting: The width of the cut seems to depend on which way the bit is faced inside the chuck.

In this image there are three sets of vertical cuts. All of the cuts in each set were made with the same bit, simply turned to face different directions.

So I printed out a little piece of paper with numbers on it and taped it to the chuck so I can record data on which way each bit needs to face to get the best cut. Hopefully that turns out to be all that there is to it, and the direction that the spring inside the chuck is facing isn't relevant.

Anyway, with that sorted out I added code to allow me to adjust the cutting depth on the fly, and it's a good thing I did as my dumb ass forgot to do the copper height measuring before starting it. So it was cutting at a height for the other side of the board. So I just told it to cut deeper, and when it was done, repeated it and told it to cut even deeper. Once I was done I added some code to hopefully make that mistake impossible in the future.

With that, I seemed to have a completed PCB. So I put some parts on it and got this:

A solder mask would be nice, but whatever I guess.

I put in a few components the wrong way around, and discovered something interesting: When you don't have plated-through holes, solder wick is very effective. It makes the lack of plated-through holes almost seem like a feature.

Anyway, the circuit is an AVR programmer I designed myself and have been using for years but built on solderless breadboard, so it's about time it's made into a PCB. I tested it and it works, so success at last I guess.