Drillbi – Detailed project overview

Technical overview

Tech stack: React 19 + TypeScript (frontend), Spring Boot 3.4 + Java 21 (backend)
State management: React Context API for global state, custom hooks for local state
Authentication: OAuth2 (Google/GitHub) + JWT tokens, Spring Security
Deployment: Docker containers, CI/CD with GitHub Actions
Testing: JUnit 5 with JaCoCo coverage reports for the backend

Architecture & integrations

Drillbi is built with a modern architecture where the frontend and backend are fully separated. The system integrates several external services to deliver a complete learning platform.

REST API: The backend exposes a RESTful API with versioning (/api/v2/)
OpenAI integration: GPT models for generating quiz questions and AI explanations
Anthropic Claude: Alternative AI model for question generation (Claude 3 Haiku)
OCR with Tesseract: Extracts text from uploaded images to create quizzes
Stripe Payments: Handles subscriptions and payments for premium features
MySQL database: Persistent storage of users, courses and questions
H2 database: In-memory database for development and testing
JPA & Hibernate: ORM for database access with automatic schema generation

Key features

AI-generated questions: Automatically create quiz questions from any text using OpenAI or Claude
Image-to-quiz: Upload images of textbooks or notes; OCR extracts the text and the AI generates questions
Adaptive repetition: Incorrectly answered questions are put back in the queue after 10–15 questions for efficient learning
AI explanations: Get detailed explanations for each question and answer with the help of AI
Multilingual support: Full support for Swedish and English with i18next
Course management: Create, edit and organise courses with custom questions
Question editing: Manually edit AI-generated questions or create your own from scratch
Statistics & progress: Real-time statistics on correct answers, error rate and progress
Premium features: Stripe integration for subscriptions with extended AI features

Screenshots

Login view where users can enter credentials or sign in with Google — 1. The login flow with support for traditional sign-in and Google auth.

Course table with several selectable quiz courses and language buttons — 2. Course overview with a language switch to start the right quiz.

Quiz question where a correct answer is highlighted in green — 3. A question session highlighting the correct answer and showing progress.

Quiz question with a wrong answer in red and an AI explanation — 4. An AI-generated explanation after a wrong answer for deeper understanding.

Form for creating a new course — 5. A simple course creator with name validation and an optional description.

Form for adding new questions to a course — 6. A question editor where you can write questions and answers and upload images.

Text-to-quiz view where text is pasted and a language is chosen — 7. Text-to-quiz generation with a live estimate of the number of questions.

Photo-to-quiz card where users can extract text from an image — 8. The photo-to-quiz flow that uploads images and extracts text.

Lessons

The backend development of Drillbi has been a deep dive into modern Java development with Spring Boot. The project gave me extensive experience in building robust REST APIs, handling complex business logic and integrating external services such as OpenAI, Anthropic Claude and Stripe. Working with Spring Security for OAuth2 authentication and JWT tokens gave me a solid understanding of secure authentication handling in a production environment.

Integrating several AI models (OpenAI GPT and Claude) required extensive work on prompt engineering, error handling and parsing JSON responses from different APIs. Handling OCR with Tesseract for image-to-text functionality was also an instructive challenge, especially around image processing and text extraction. The Stripe integration taught me how to handle webhooks, payment flows and subscription management securely.