Overview
Noory is an AI-powered book recommendation app designed to help children in Arabic-speaking countries discover engaging books and audiobooks that match their interests, age, and learning stage. As one of the region’s most prominent edutainment platforms, Noory blends education with entertainment, making reading feel like play.
Blink22 partnered with Noory to design and build a custom machine learning recommendation engine that transforms vast amounts of behavioral data into highly personalized suggestions, turning the children’s reading app into a smart companion for kids.
Technologies Used
Technologies Used- TensorFlow
- BigQuery
- Apache Beam
Category
- AI
- Education
- Entertainment
- Edutainment
Platforms
- iOS
- Android
App Links
Initial Requirements
Noory set out to solve a common challenge among parents: getting kids excited about reading. But unlike traditional apps, they aimed to use machine learning and data science to make recommendations not only accurate but also meaningful and fun.
To bring this vision to life, they needed a technology partner with deep experience in ML model development and training, real-time data processing at scale, and seamless integration into mobile apps.
Blink22 was chosen to lead the full lifecycle of Noory’s recommendation engine, from data preparation and model tuning to backend deployment and system integration.
Challenge
Building a recommendation system for children presented a unique set of challenges:
Extracting and transforming unstructured event data from user activity.
Designing ML models that could detect age-appropriate interests, learning styles, and engagement patterns.
Delivering high-accuracy recommendations in real-time across books and audiobooks.
Ensuring recommendations remained safe, culturally relevant, and adaptable over time.
Achieving this required not only algorithmic precision but also a system architecture that could scale and adapt quickly to feedback.
Approach
Blink22 brought together a team of machine learning engineers and data scientists to build a powerful AI pipeline tailored for Noory’s goals:
Data Engineering & Preprocessing
Ingested and cleaned large volumes of user interaction data.
Applied event tracking pipelines using Apache Beam and stored data in BigQuery for analysis.
Model Training & Tuning
Designed and trained ML models using TensorFlow and the Python ML stack (scikit-learn, pandas, numpy).
Iteratively tuned models to improve recommendation accuracy for multiple content types (books, audiobooks).
System Integration
Integrated the trained models into Noory’s backend and exposed intelligent book suggestions to the mobile frontend.
Enabled real-time, context-aware suggestions based on user behavior, preferences, and feedback.
Solution
The result was a highly personalized and scalable book discovery engine:
Recommends the right content at the right time, tailored to each child’s reading journey.
Works seamlessly across multiple formats: print books, audio content, and more.
Supports continuous learning, so the system evolves as kids grow and their interests change.
Outcome
Today, Noory stands as a benchmark in Arabic-language edtech, bridging the gap between AI-driven learning tools and early childhood education. The results:
100K+ downloads on Google Play with a 4.3★ rating.
Kids now enjoy personalized recommendations that keep them engaged and curious.
Parents can explore audiobooks tailored to their child’s development stage.
The app’s machine learning engine grows smarter with every interaction, enhancing user engagement over time.