developing a Monocular SLAM system

Hello everyone! For a uni project I want to develop a very basic monocular SLAM system.

Around the net I could find plenty of papers, state-of-the art system and new DL approaches but little about how to develop a very basic system from scratch.

I'm pretty handy with numpy and OpenCV and have good real-world experience with feature extraction/detection as well as position estimation and real time visual tasks.

Could you point me to some good reads or somewhere to …

art basic computervision good hello numpy opencv project slam state world