Description

In this training you will learn how to set up a Retrieval-Augmented Generation (RAG) system that allows you to have large language models (LLMs) answer based on your own documents, knowledge sources or internal systems. By combining document search with text generation you create AI applications that are more up-to-date, reliable and domain-specific than generic language models alone.

You will discover how to convert documents to embeddings, store them in a vector database and link them to an LLM via retrieval. We work with tools such as LangChain , LlamaIndex , Hugging Face Transformers and vector databases such as FAISS , Weaviate , Pinecone or Chroma . You will also learn how to use PDFs, Word files, web pages or internal reports as input and how to build a smart interface or API on top of your RAG pipeline.

Learning objectives

• Understanding what a RAG architecture is and when to use it (vs. fine-tuning or prompt engineering)

• Process and chunk documents in a way that fits your use case

• Generating Embeddings with Models via Hugging Face or OpenAI

• Using vector databases such as FAISS, Weaviate or Pinecone for semantic searches

• Apply metadata filtering, ranking and hybrid search for more accurate retrieval

• Building RAG pipelines with LangChain or LlamaIndex

• Integrate LLMs via Hugging Face Transformers, OpenAI API or Anthropic API

• Building a working application with a front-end (e.g. Streamlit) or REST API

Approach and working methods

The training is completely hands-on. You build a working RAG solution step by step: from document processing to integration with a language model and a working interface. We cover all layers of the architecture and you learn how to make choices based on data volume, user needs and scalability.

Examples of applications you can build:

• A legal or policy assistant based on your own documents

• An HR bot that provides answers based on personnel information and regulations

• An IT search assistant that answers helpdesk questions based on manuals and internal documentation

• A project assistant who interprets and summarizes reports, minutes and decisions

You work with:

• Python and frameworks such as LangChain or LlamaIndex

• Vector databases : FAISS, Chroma, Weaviate, Pinecone

• Embeddings of OpenAI or Hugging Face (e.g. BGE, Instructor, MiniLM)

• LLMs via OpenAI, Hugging Face Hub or own hosted models

• Frontend/API via Streamlit, FastAPI or Flask

For whom

This course is intended for AI engineers, ML specialists, developers and data scientists who want to develop smart AI applications based on their own documents or knowledge sources. Experience with Python and basic knowledge of LLMs and vector search is recommended.

Interested in this training?

Feel free to contact us. We are happy to tailor the training to your specific use case, type of documents and technical preferences.