Dersify — AI-Native Personalized Learning

AI-native education product designed around persistent learner modeling, diagnostic learning paths, retrieval practice, spaced repetition, and adaptive AI feedback.

AIPersonalized LearningNestJSNext.jsPostgreSQLPGVectorAnthropicOpenAI

Engineering Decisions

Designed the product around a persistent learner model rather than one-off AI tutoring interactions.

Combined diagnostic learning maps, retrieval practice, spaced repetition, and error-driven adaptation into the product architecture.

Designed a NestJS / Next.js / PostgreSQL architecture with AI provider abstraction, queues, vector search, and asynchronous learning workflows.

Structured the platform for long-term learner state and adaptive learning rather than stateless prompt-response sessions.

Architecture

LearnerDiagnostic ModelLearning MapAI EngineLearner StateSpaced RepetitionAdaptive Feedback

Data & AI Flow

Learner performance updates a persistent learner model, which drives future content selection, retrieval practice, scheduling, and AI feedback.

Engineering Tradeoff

Chose persistent learner modeling over generic chatbot tutoring because personalization requires longitudinal state.

Proof Points

ArchitectureTradeoffScale