Skip to content
EdTech

Adaptive Problem Recommender

Recommend the next problem based on a model of what the student knows.

Advanced24h2–3 people16/20 fit
TL;DR

Build Adaptive Problem Recommender for online practice platforms in ~24h with 2–3 people. Stack: Python + FastAPI + Next.js.

Why it wins: Real learning science.

Build flow

01 · Problem
Fixed problem sets bore strong students and lose weak ones.
02 · Solution
Model skill mastery with knowledge tracing, recommend problems in the right difficulty band and show teachers mastery heatmaps.
03 · Build
  1. 1Tag problems by skill
  2. 2Implement tracing
  3. 3Recommend
  4. 4Visualise
04 · Demo
Answer a few problems and watch recommendations adapt.

MVP checklist

  • Skill model
  • Recommendations
  • Mastery heatmap
  • Teacher view

Tech stack

TechRole
PythonBackend / scripting
FastAPIExternal data
Next.jsFrontend + API routes

Related ideas

More EdTech

  • EdTech

    Textbook Tutor (RAG)

    Answers questions only from your own textbook, with page citations

    Intermediate24h1–3
    Next.jsRAGVector DB
    17/20

  • EdTech

    Socratic Math Tutor

    A tutor that asks guiding questions instead of giving answers

    Intermediate14h1–2
    LLMKaTeXNext.js
    17/20

  • EdTech

    Concept Map Generator

    Turn notes or a chapter into an editable concept map

    Intermediate12h1–2
    LLMExcalidrawNext.js
    17/20