Say hello WRITING
Notes on growing
AI, teaching, and building products — written down so I actually understand them.
27 Aug 2026 Where you cut the text Chunking strategies for RAG: what the benchmarks show about chunk size, overlap and semantic splitting — and how to choose and measure your own. 14 Aug 2026 When the model makes things up You can't delete LLM hallucinations, but you can stack defences. A practitioner's playbook — grounding, prompts, refusal, verification, and measuring it. 9 Aug 2026 Three ways to give a model your knowledge Confused between RAG, fine-tuning and long context? A vendor-neutral decision framework — with cost, latency and a real worked example. 9 Aug 2026 How do you know your LLM app works? Stop shipping LLM features on vibes. A from-scratch guide to evals — golden sets, the metrics that matter, LLM-as-judge, and a real example. 21 Jul 2026 Scout: an agent that decides where to look AI agents explained from first principles — tool calling, the decide-act-observe loop and knowing when to stop — via a real agent that cites its sources. 20 Jul 2026 The Writing Bed: answering from my own words RAG explained from scratch — embeddings, chunking, vector search and grounded answers with sources — built through a tool that answers from my blog posts. 19 Jul 2026 Growing Sprout: a digital twin that knows its limits How to build an LLM chatbot that stays accurate — system prompts, grounding, streaming and guardrails. The hard part is teaching it to refuse. 15 Jul 2026 Rebuilding Usama's corner of the internet Why I retired my old FastAPI portfolio for a static site with a proper brand — and what 'helping people and products grow' means as a design brief.
No writing matches “”. Try another term.
Let's grow something together.
Consulting, teaching, speaking, or a product idea that needs an AI brain — my inbox is open.
