Content

What is RAG? (60-sec explainer)

Retrieval-Augmented Generation explained in under a minute — why LLMs hallucinate and how grounding in a knowledge base fixes it.

How Attention Works in Transformers

Visual walkthrough of the self-attention mechanism — queries, keys, values, and why the dot-product matters.

Vector Databases in 60 Seconds

What vector databases are, how similarity search works, and when you actually need one vs. a regular database.

The Agent Loop Explained

Perceive → plan → act → observe. How modern AI agents work, from ReAct to tool-calling, in a short visual format.

ColPali: Document Retrieval with Vision Language Models

Deep dive into the ColPali paper — how page-level visual embeddings replace text extraction for PDF search, and when to use it.

Build a RAG Pipeline from Scratch

End-to-end walkthrough: chunking, embedding, vector store, retrieval, and generation. Using LangChain and OpenAI.

Context Engineering > Prompt Engineering

Why structuring what goes into the context window matters more than tweaking prompt phrasing — and how to think about it systematically.

What Agentic SDLC Actually Looks Like

Breaking down how AI agents fit into real software delivery workflows — not hype, but the actual patterns we use at Heizen.