RAG & LLM Engineering: Build AI Applications from Scratch

This course contains the use of artificial intelligence.

Learn the foundations of modern LLM engineering and Retrieval-Augmented Generation (RAG) and discover how intelligent AI applications are built from the ground up.

This course begins with the AI engineering roadmap and introduces the architecture behind RAG systems. You’ll understand why RAG is used to improve LLM responses, how it helps address hallucinations, and how native Python can be used to build flexible AI applications without depending heavily on frameworks.

You’ll then explore how to work effectively with LLM APIs, including important concepts such as tokens, temperature, Top-P, context windows, memory, context caching, and API costs. Through practical demonstrations, you’ll build a robust Python API wrapper and implement real-time streaming responses.

Next, you’ll learn the foundations of embeddings and vector search. You’ll understand how semantic meaning is represented in vector space, compare similarity techniques such as cosine and Euclidean distance, generate embeddings with Python, and build your own similarity search function.

The course also takes you inside vector databases, covering indexing concepts such as HNSW and IVF and comparing popular solutions including Pinecone, Qdrant, and ChromaDB. You’ll learn how to provision a cloud vector database and perform essential database operations using Python.

Finally, you’ll explore one of the most important parts of a RAG system: data ingestion. You’ll learn professional chunking strategies, work with PDFs, tables, and Markdown, enrich data with metadata, create a custom semantic text splitter, and build an automated ingestion pipeline.

By the end of this course, you’ll have a strong practical foundation in LLM APIs, embeddings, vector databases, data processing, and RAG architecture, preparing you to build more advanced and reliable AI-powered applications.