The study focuses on the architectural and methodological principles of designing a cross-platform information system for organizing learning and automated testing in the context of digital transformation of education. It examines modern educational information systems, including universal learning management systems (LMS) and assessment-oriented systems, analyzing their systemic limitations such as functional redundancy, architectural inflexibility, and weak alignment with pedagogical methodologies. Particular attention is given to the comparative analysis of monolithic, microservice, and service-modular architectures as applied to educational systems, as well as to methodological adaptation of testing, including support for various assignment types and adaptive knowledge assessment. The study is based on methods of systems and comparative analysis, architectural modeling, and functional design of information system components. The study confirms the existence of a methodological gap between pedagogical requirements and technological capabilities of existing platforms and justifies the service-modular architecture as the optimal model for building cross-platform educational systems. The author's key contribution is the development of the concept of transitioning from a "ready-made platform" to a "modular educational environment builder," where each module is designed as an autonomous service allowing flexible composition in accordance with specific pedagogical objectives. The novelty of the study lies in the development of a typology of architectural models and criteria for their comparative evaluation. Furthermore, the originality of the findings is reflected in the justification of the service-modular approach as a balanced solution that addresses both technological requirements such as scalability and cross-platform compatibility, and methodological requirements of variability in learning forms, adaptive testing, and analytical support for the educational process.
Vaganov et al. (Sun,) studied this question.