Adaptive Learning Platform — An AI-powered adaptive learning platform for K-12 students.
Livemobile2024

Adaptive Learning Platform

Client

EdTech · K-12 Education

Year

2024

Category

mobile

Tech Stack
FlutterFirebasePythonTensorFlowGCP
The Challenge

What they needed

The platform needed one Flutter app to serve K-12 students spanning early readers to teenagers, each with a different pace and learning style, across a curriculum of 500+ topics. A single fixed sequence doesn't hold at that range — a student can be ahead in one subject and behind in another, and rigid pacing kills engagement fast. The team had to build a recommendation layer that continuously reassesses mastery per topic and reroutes content automatically. Mobile added its own constraints: the app had to stay fast and responsive across a wide range of devices, and the underlying ML pipeline had to turn usage signals into updated recommendations quickly enough to feel current as the platform scaled toward 75,000+ active students.

Our Solution

What we built

We built the student-facing app in Flutter for a single codebase across iOS and Android, covering all 500+ curriculum topics without maintaining separate native apps. Firebase handles auth and real-time data sync, so every activity a student completes updates their progress instantly rather than on a delay. The recommendation engine runs in Python, with TensorFlow models trained on student interaction data to score topic mastery and select the next best piece of content per student. GCP hosts training and serving infrastructure, scaling inference as active users grew. A gamification layer sits on top of the recommendation engine to sustain engagement. Progress tracking was built as two distinct views reading off the same Firebase data layer: a simplified snapshot for parents, and topic-level detail for teachers.

Key Decisions

How we approached it

01

Built the app once in Flutter instead of two native codebases, making it feasible to cover 500+ curriculum topics without doubling mobile engineering effort.

02

Modeled personalization as continuous per-topic mastery scoring in TensorFlow rather than fixed grade-level pacing, so a student's path shifts as performance data comes in.

03

Used Firebase for real-time sync so parent and teacher dashboards reflect a student's activity the moment it happens, not on a batch delay.

04

Split progress tracking into two purpose-built views on the same data — a simplified summary for parents, topic-by-topic detail for teachers — instead of one generic report for both.

Results

The impact

75000+

Active Students

40%

Learning Gain

4.9

App Rating

"Our students are more engaged than ever. The AI personalization is incredible."

Head of Product

EdTech platform

App · platform · database · or AI

Tell us what
to build.

Bring us the one everyone said was too hard. Free, no-obligation reply within 24 hours — NDA on request.