I think you’ve gotten a bunch of good answers already in this thread. I’ll try and summarize some of them.
Kanji characters have multiple ways of being read, in different categories. For the purpose of using WaniKani, we focus on the “main” two: On’yomi (the “Chinese” reading), kun’yomi (native reading). On’yomi refers to pronunciations borrowed from the Chinese language, from many different periods. Kun’yomi refers to pronunciations created within Japan, for Japan (presumably).
A single kanji character can have one or more of each. I think the highest amount for a single character is 12 or something.
On’yomi is most frequently found in words that are composed of two or more kanji.
Kun’yomi is most frequently found in words that are composed of a single kanji, or that can have hiragana/katakana in them.
WaniKani will try and first teach you what it thinks is the single most, or two most important, pronunciations for the character. You may learn the rest through vocabulary.
There is no way to know in advance what word uses which pronunciation unless the kanji has a single, easy pronunciation. Those are good words.
And yes, there are words which have exceptional ways to read them, with no warning beforehand.
This video may be fun to watch! It has some more info about why kanji have multiple readings, multiple times: https://www.youtube.com/watch?v=CF3MRMBjd20