Microservices-Based Learning Platforms for Adaptive Education

Abstract

Adaptive education platforms -- which personalise learning pathways, content difficulty, and instructional modality based on individual learner models -- require software architectures that support the simultaneous operation of heterogeneous AI components (knowledge tracing models, recommendation engines, adaptive assessment systems, and real-time feedback generators) alongside traditional LMS functions. Monolithic LMS architectures cannot support this AI-augmented adaptive stack without significant coupling between educational AI services and core platform functionality. Microservices decomposition enables independent development, deployment, and scaling of each adaptive component while maintaining integration via well-defined APIs. This paper proposes the Microservices Adaptive Learning Architecture Framework (MALAF), a systematic design and evaluation of a microservices architecture specifically optimised for adaptive education platforms. MALAF decomposes the adaptive learning platform into 14 microservices across five functional domains -- learner modelling, content management, adaptive delivery, assessment, and analytics -- and evaluates the architecture on three quality attributes: adaptivity responsiveness, scalability, and developer agility, across 18-month deployment at two adaptive learning institutions. Key results: MALAF achieves 94 ms mean adaptive recommendation latency (< 100 ms SLA target); independent service deployment reduces new AI feature delivery time from 6.4 weeks (monolith) to 1.2 weeks; 99.96% availability at 40,000 concurrent users. The framework contributes a reusable microservices blueprint for adaptive education platform architects.