How to build an adaptive learning/recommendation system for a question bank? [D]

Hey! Can you tell me how you would go about building a recommendation engine for our question bank?

The idea is that it understands a student’s strengths and weaknesses and recommends questions accordingly — more questions around the areas they’re weak in, but without making them so difficult that they feel demotivated.

I also want it to occasionally bring back questions from older topics to check whether the student has forgotten something. Based on how they perform, it could then decide whether to recommend more questions from that topic or move on.

Basically, the goal is for the recommendation engine to continuously understand where the student is struggling and use that to help them become better at problem-solving over time.

I was learning some basics of AI/ML and this question came to my mind, so I was just curious — do you have any idea how something like this could be built?