It really depends. Were you wanna deep ends.
If you want to go towards gfx, you will need to study up on the basics (matrices, cameras, transformations etc.) and then you should dive into ray-trace lighting (pbr) because that is the future. Its also nice to learn glsl, just to see the mind boggling trickery used today to make games beautiful.
If you want to get deeper into engine programming, i recommend reading into a existing one. DO NOT START TO PROGRAM YOUR OWN ENGINE. Yes, you learn alot, but no, you will not have a game.
There are lots of interesting subdomains in engine:
- Pathfinding
Optimizations (Producing performant, readable code is hard- VERY HARD)
AI
Features
I personally recommend picking up bugs in open source engines and contribute by fixing them.
AIs:
There is a little rift going through the AI-World right now.
On the one half, are the "old" AIs that consist of hardcoded modules, tailored to one game or at best a game-category, where they excel. If the game changes to much, they can decay. So they need maintenance.
Some of the newer generations used a configurable AI (Shard) that can be customized towards the game and thus does not need so much maintenance.
Then there are simple script AIs that basically are just small scripts putting pressure on the players.
In the new AI-World, you have things like Alphastar.
https://deepmind.com/blog/article/alpha ... arcraft-ii
Ruthlessly efficient killermachines, who do not have any code, just input, a trained NN and some output
If you want to venture towards GameDesign, you will have alot to read and play, here its more important