I believe it depends on the era and system, but there were various APL codepages (i.e. definitions for the upper 128 characters) for both EBCDIC and ASCII.

https://en.wikipedia.org/wiki/Digital_encoding_of_APL_symbol...

In the very earliest IBM Selectric teletype-based systems, some APL symbols were constructed by entering one character, hitting backspace, and overstriking a second character. For instance, ⍋ is | overstruck on ∆. It's why a lot of APL symbols look like that.

https://aplwiki.com/wiki/Overstrike

Early on, the selectric typewriter thing had a spherical ball that could rotate to stamp the characters. So when you typed a key the IBM hardware would type a character on a piece of paper exactly like a typewriter and also the IBM computer would keep track of this and when you ran the expression it would calculate the result and print that out as well. You can see videos of this.

Custom encodings, as was standard (or, well, mandatory) before Unicode (1991). Hell, Dyalog APL to this day supports its classic 1-byte-per-char encoding (not even ASCII-compatible! Nor EBCDIC!) in addition to Unicode.

Looks like the APL chars were added in Uncicode 1.1 (1993), two years after 1.0, which is quick enough.