Johanna AI – Project Overview

Johanna AI is an assistant designed to support current and aspiring Psychology students.
The full solution is composed of three independent applications that communicate with one another:

---

1. Web Client

• Repository: https://github.com/bispo-daniel/React_Johanna
• Stack: Vite + React 18, TypeScript, and Tailwind CSS
• Role: delivers the web interface where users—either "super user" or "commom user" — sign in, chat with Johanna AI, view answers in real time, and (for "super user") approve pending registrations submitted by common users.
• Integrations
  • REST API (Java) for authentication and persistent data
  • WebSocket for streaming model responses

For complete details on scripts, dependencies, and local setup, see the package.json in this repository.

---

2. WebSocket Server

• Repository: https://github.com/bispo-daniel/johanna_openai
• Stack: Node.js, Express, Socket.io, and TypeScript
• Role: receives messages from the front-end, forwards them to the fine-tuned OpenAI model, and streams the responses back to the client
• Technical highlights
  • Payload validation with Zod
  • JWT support to identify authenticated users
  • Sensitive variables loaded via dotenv

---

3. REST API

• Repository: https://github.com/DevRodrigoSantana/Smart-teaching-assistant-Johana
• Stack: Spring Boot 3 (Java 17), Hibernate JPA, Spring Security, and MySQL
• Role: business core — manages users, JWT authentication, data persistence, and domain rules
• Key dependencies
  • spring-boot-starter-web and spring-boot-starter-data-jpa for the web layer and ORM
  • spring-boot-starter-security with jjwt for authentication
  • ModelMapper for DTO ↔ entity mapping

---

How the layers fit together

1. The Web Client sends login requests and other REST operations to the API.
2. Once authenticated, the browser opens a WebSocket channel to chat with Johanna AI.
3. The WebSocket Server connects to OpenAI, receives streamed responses, and pushes them back to the front-end.
4. User data, sessions, and history are stored by the API in a MySQL database.

---

For questions or contributions, visit each repository above and consult the relevant configuration files (package.json, pom.xml).