Skip to content

Loading...

Automation 5 ay AI Researcher & Full-Stack Developer

Adaptive Learning Platform

EdTech platform offering AI-powered personalized learning experiences tailored to each student. Dynamic curriculum optimization with Reinforcement Learning.

%35
Success Rate Increase
%50
Churn Reduction
200K+
Active Students
%25
Learning Time Reduction

Challenge

Creating personalized learning paths for 200,000+ students, accurately modeling learning patterns, and optimizing the 'boredom threshold' were the main challenges.

Solution

Continuously tracked student knowledge levels with Deep Knowledge Tracing model. Maximized engagement with dynamic content selection using Multi-Armed Bandit algorithm. Increased knowledge retention with Spaced Repetition principles.

Highlights

1

Student modeling with Deep Knowledge Tracing

2

Dynamic content optimization with Multi-Armed Bandit

3

Spaced Repetition-based review system

4

Scalable architecture (200K+ concurrent users)

Technology Stack

Python
TensorFlow
scikit-learn
Next.js
Node.js
MongoDB
AWS

About the Project

Developed for an education platform with 200,000+ active students, this AI system delivers content tailored to each student's learning pace and style.

Technical Details

  • Multi-Armed Bandit for content selection optimization
  • Knowledge Tracing model (DKT - Deep Knowledge Tracing)
  • Student profiling and clustering
  • Automated question difficulty level determination
  • Performance prediction model
  • Results

    Achieved 35% increase in student success rates, 50% reduction in platform churn rate, and 25% decrease in average learning time.

    Adaptive Learning Platform | Şükrü Yusuf KAYA