StudyMate – AI-Powered Adaptive Learning and Exam Intelligence Platform

Adaptive Learning & Exam Intelligence Platform for Cram Schools

12 members × 10 months

PM / BA / Developers / QC

Requirements Definition / UI-UX Design / Development / Testing / Deployment / Maintenance & Operations

• LightGBM (score prediction)

• Bayesian Knowledge Tracing (BKT – mastery estimation)

• Item Response Theory (IRT – difficulty optimization)

• GPT-4o

• LangChain / Qdrant

• Azure OpenAI / text-embedding-3-large

• Python (FastAPI)

• Neo4j (Knowledge Graph)

• PostgreSQL / Redis / Celery

• Flutter 3.x / React 18 / TypeScript

• Recharts (data visualization)

• Apache Airflow / dbt

• BigQuery

• GCP (GKE / Cloud Run / Vertex AI)

• MLflow

• GitHub Actions

• Predicts national exam scores (Common Test)

• Visualizes subject-wise score trends

• Supports university recommendation and admissions planning

• Real-time learning progress tracking

• Accuracy rate and study time analysis

• Weakness area visualization

• Automated grading for essay-style questions

• Simplified score input workflow

• Auto-generated performance reports

• Weakness analysis reports per student

• Recommended personalized teaching plans

• Class-level understanding analytics

• Deployed to 8,000 students within 4 months

• AI grading system processes 1,500 answers/day, reducing grading workload by 65%

• Score prediction achieved MAE ±42 points, used for admission guidance

• Improved learning efficiency through personalized weakness-based study

• Parent satisfaction increased from 3.8 → 4.7

• 68% of nighttime student questions resolved by AI automatically

• Reduced grading workload by 65%, reallocating time to teaching

• Early Warning System detects dropout risk 6–8 weeks in advance

• Teacher onboarding time reduced from 4 weeks → 1 week

• Exam success rate increased by +27%

• Student retention improved from 54% → 76% (+41%)

• Enrollment conversion increased by +23%

• Highly praised for deep understanding of Japanese education and exam systems

• Strong evaluation of Common Test alignment and subjective question design

• Recognized for student psychology-aware learning design

• Expansion under consideration for Eiken & TOEFL preparation markets