Large Language Models
What Is a Large Language Model (LLM)?
How LLMs actually generate text, what tokens and context windows really mean, and why models hallucinate — explained without the jargon.
In-depth, practical writing on Generative AI — organized around the topics that actually matter for building real applications: LLMs, RAG, AI agents, vector databases, prompt engineering, evaluation, and production AI.
Topic cluster
Large Language Models
How LLMs actually generate text, what tokens and context windows really mean, and why models hallucinate — explained without the jargon.
Topic cluster
Retrieval-Augmented Generation (RAG)
Why RAG exists, how a retrieval pipeline actually works, and how it differs from fine-tuning — with the "open-book exam" analogy that makes it click.
Topic cluster
Vector Databases
How vector databases power semantic search: embeddings, similarity measures, ANN indexes, and choosing the right one for RAG.
Topic cluster
AI Agents
The mechanics behind AI agents and Agentic AI: tool calling, the agent loop, and why reliability is the hard part.
Topic cluster
Prompt Engineering
Zero-shot to few-shot prompting, structured outputs, instruction hierarchy, and how to evaluate prompts like code — not guesswork.
Topic cluster
LLM Evaluation
Why "it looks right" isn't evaluation, and how to actually test LLM and RAG systems with metrics, LLM-as-judge, and a real test set.
Topic cluster
Production AI
The unglamorous engineering that turns an AI demo into a reliable product: architecture, caching, cost control, logging, and security.