☕️ Welcome to this workshop to build your own Java AI Agent using Retrieval Augmented Generation.
🏆 It leverages the best of open-source for fast implementation of the RAG pattern for production quality applications.
🤩 The completed RAG ChatBot will demonstrate how your AI Agent can do
- LLM and Prompt Engineering
- Conversational Memory
- Vector Similarity Searching and Dense Passage Retrieval
- Transform, chunk, and vectorise unstructured files like PDFs
- Caching of LLM responses for latency and cost performance
- Reranking of search results
- Vector calculations using JVector
- Online searching using the Tavily service
- Hybrid Searching
- Closed Loop Feedback System
- LLM Function Calling
- Time Series Vector Similarity Searching
♻️ This workshop uses Java 21, Spring AI and Vaadin for the UI. The use of Spring and Vaadin is minimal, the code is intended to be re-used in other frameworks.
👩 It is CQL compatbile with Apache Cassandra® 5.0 and AstraDB Vector. Database schemas and data models are intentionally flexible so the concepts in the workshop can be retrofitted to your needs and your production.
🙇 The workshop will use the services: OpenAI, Tavily, and AstraDB. You will need accounts and api keys for each of these.
🌴 Each step in the workshop is a separate branch in this repository, you will need to be familiar with git switching between branches.
💪🏽 To move on to read the requirements setup step, do the following:
git switch workshop-intro-requirements
- Java 21 (or beyond)
- Command line
gitand git knowledge - DataStax Astra DB (you can sign up through your Github account)
- OpenAI account (you can sign up through your Github account)
- Tavily account (you can sign up through your Github account)
- An editor of your choice
All work is copyrighted to DataStax, Inc