""We needed a system that could actually teach — not a shinier Moodle." — Dr. Priya Nair, Founder"
The Challenge
EduSpark had spent 18 months customising Moodle. The result: a 34% course completion rate and instructors who had zero visibility into where learners were dropping off.
Our Approach
An adaptive learning engine constructing a knowledge graph per student, continuously recommending optimal next content based on performance and interaction patterns.
- OpenAI embeddings tagging 50,000 content items across a curriculum knowledge graph
- Spaced repetition algorithm for long-term memory retention optimisation
- LangChain AI tutor answering questions in full curriculum context
- Instructor dashboard with real-time confusion signals and dropout risk scores
Results
Course completion: 34% to 85%. Average session length up 2.5×. 40 university partners and 100,000+ active students in year one.
Technologies Used
React.js
FastAPI
PostgreSQL
Redis
OpenAI
LangChain
AWS
Swift
Kotlin