OnboardEase - Dynamic Learning Assistant with Langchain4j & Spring AI 🦜️🔗 + 🍃🤖️

OnboardEase: A dynamic learning assistant designed to simplify the onboarding and training process for new hires. Users can upload documents or enter URLs for training materials, and the application provides a platform for asking training-related queries. Built with Spring Boot, Langchain4j and Spring AI.

💡 This application is not business oriented and my focus is mostly on technical part, I just want to play with LLM/RAG chatbot assistant by building a sample app from scratch with two main libraries in java ecosystem : Spring AI 🍃 and Langchain4j 🦜️.

🌀 This Application is in-progress and I will add new features over time. 🌀

Application Screenshot

Features

• ✅ Choose between two libraries for chatbot assistant : Langchain4j or Spring AI
• ✅️ Upload documents or enter URLs for training materials
• ✅️ Ingest documents and give context to the chatbot
• ✅️ Ask training-related queries

Tech Stack

• 🍃️ Spring Boot
• 🦜️ Langchain4j
• 🤖️ Spring AI
• 📦️ Maven
• 🐳️ Docker
• 🐘️ PostgreDB
• 🔍 PgVector (Postgres extension for similarity search)
• 🧠 Openai
• 🌐️ Thymeleaf

Setup & Installation

• 1️⃣ Clone the project on your local machine

  git clone https://github.com/miliariadnane/onboardease-rag-training-assistant.git

• 2️⃣ Install dependencies

  mvn clean install

• 3️⃣ Run docker-compose file :

  docker-compose up -d

• 4️⃣ Open PgAdmin UI and create two databases : vector_store and langchain_vector_store
  • 4️⃣- 1️⃣ For vector_store database, create an extension (do this once in each database where you want to use it) :

    CREATE EXTENSION vector;

• 5️⃣ Configure your openai API key in application.properties file or in your environment variables
• 6️⃣ Run the application

  mvn spring-boot:run

• 🚀️ Open your browser and navigate to http://localhost:8080 (Enjoy the app 🎉️)

How RAG Works under the hood ?

• 1️⃣ Ingest domain data and give context to the chatbot.
• 2️⃣ Reader to retrieve text from the documents.
• 3️⃣ Splitter to split the text into chunks.
• 4️⃣ Embedding to convert the text into vectors.
• 5️⃣ Store the vectors in the vector store database.
• 6️⃣ User ask the question to the chatbot.
• 7️⃣ Splitter to split the question into chunks.
• 8️⃣ Embedding to convert the question into vectors.
• 9️⃣ Similarity search to find the most relevant chunks between the question and the documents ingested.
• 🔁️ Loop through the chunks and return the most relevant answer to the user.
• 🎉️ Return the answer to the user.

Diagram of the Application

Roadmap (In Progress)

• 🚧 Fix AI text embedding issue (Spring AI)
• 🚧 Implement similarity search to find the most relevant answer to the user (Spring AI)
• 🚧 Add system message prompt (Spring AI)

Support

• ⭐️️ Star this repository if you find it useful.
• 🐛️ If you find a bug, raise an issue or fix it and send a pull request.
• 📢️ If you have any feature requests, raise an issue or send a pull request.
• 🤲 If you have a moment, don't forget to make a duaa for me and my parents.