EdTech

LearnPulse

Adaptive Learning at 200K Student Scale

ClientLearnPulse Education Inc.
Duration13 months
Team9 engineers + 2 learning scientists
Year2024
🎓

📋 Overview

LearnPulse wanted to move beyond static online courses. We built a personalised learning platform that adapts in real time to each student's pace, knowledge gaps, and learning style — serving 200K+ concurrent students across K–12 and university curricula in 6 languages.

⚠️ The Challenge

Their existing LMS was a WordPress plugin serving pre-recorded videos. Completion rates were 18%. Students with different starting knowledge levels were getting the same content. Teachers had no visibility into where students were struggling. The platform was unusable on mobile — where 70% of their users were.

💡 Our Solution

We built a knowledge graph of 50K+ learning objectives with prerequisite mapping. A Bayesian knowledge tracing model updates each student's mastery level after every interaction. Content is dynamically assembled — videos, interactive exercises, and quizzes — tuned to the student's zone of proximal development. Teachers get a real-time class heatmap.

Results That Speak

🎯73%Course Completion Rate (was 18%)
👥200K+Active Learners
🌍6Languages Supported
🚀40%Faster Learning Outcomes
92%Student Satisfaction Score
🧠50K+Learning Objectives Mapped

Key Features

🧠

Bayesian Knowledge Tracing

Real-time mastery estimation updates after every quiz answer, video pause, or exercise attempt.

🗺️

Knowledge Graph

50K+ learning objectives connected by prerequisite relationships, enabling true adaptive sequencing.

📊

Teacher Command Centre

Live class heatmap showing per-student mastery, at-risk flags, and suggested interventions.

🌍

6-Language Support

Full content localisation with RTL support, auto-translated subtitles, and locale-aware exercises.

📱

Offline Mobile Learning

Smart pre-cache downloads next 3 lessons when on WiFi — works fully offline on mobile.

🏆

Gamification Engine

Streaks, badges, leaderboards, and XP — designed with learning scientists to boost intrinsic motivation.

Technology Stack

AI / Personalisation

PythonPyTorchBayesian KTKnowledge GraphsNeo4j

Backend

Node.jsFastAPIPostgreSQLRedisWebSockets

Frontend

Next.jsTypeScriptFramer MotionVideo.jsD3.js

Mobile

React NativeExpoOffline Content CachePush Notifications

Infrastructure

AWSCloudFrontS3Lambda@EdgeDatadog

Project Timeline

01

Learning Design & Research

6 weeks

Learning scientist engagement, knowledge graph taxonomy, BKT model research, curriculum mapping.

02

AI Personalisation Engine

14 weeks

BKT model implementation, Neo4j knowledge graph, recommendation engine, A/B testing framework.

03

Content Platform

10 weeks

Video delivery pipeline, interactive exercise engine, quiz system, content authoring tools.

04

Web Platform

8 weeks

Next.js learner dashboard, teacher command centre, D3 visualisations, real-time WebSocket updates.

05

Mobile App

8 weeks

React Native app, offline content caching, push notifications, gamification UI.

06

Scale & Internationalisation

11 weeks

6-language localisation, CDN optimisation, load testing to 300K concurrent, rollout.

Going from 18% to 73% completion rate is the kind of result you hope for but rarely see. The personalisation engine genuinely feels like a one-on-one tutor. Teachers tell us they finally feel like they understand their students.
AK
Ananya Krishnan
Chief Product Officer, LearnPulse
★★★★★

Ready to build something like this?

Tell us about your project and we'll put together a tailored proposal within 24 hours.

START YOUR PROJECT →
NEXT CASE STUDY

DataVault

Read the story