15Oct
Databases · 6 min readNewB-Trees or LSM-Trees: Your Database Already ChoseEvery database picks a side in the same trade — quick to write, or quick to read. Almost every performance surprise you have had follows from which side it picked.The things I got wrong, written down.
Long-form essays on distributed systems, retrieval and AI plumbing. Each one starts from a bug I shipped or a belief I had to drop.
11 matches
October 2026
3 essays
8Oct
Distributed Systems · 5 min readMobile Apps Are Distributed Systems With Terrible NetworksA phone is a copy of your data that vanishes without warning, holds changes for hours, and comes back disagreeing with everyone else. Most mobile bugs are really that problem, unlabelled.
1Oct
Databases · 6 min readWhat I Got Wrong About Document ModellingPicking MongoDB takes five minutes. Deciding what one record should contain takes months to get wrong and is painful to undo.
September 2026
4 essays24Sep
Distributed Systems · 5 min readYour Retry Made It WorseTrying again after a failure is the most obvious safety net there is, and the easiest one to build backwards. It hinges on one question: did the first attempt actually fail?
17Sep
Agents · 5 min readMCP Is a Protocol, Not a Plugin SystemThe Model Context Protocol looks like a way to hand an AI some plugins. Treat it that way and you get tools that shine in a demo and go round in circles in real use.
10Sep
LLM Systems · 6 min readLocal-First AI Is a Systems Problem, Not a Model ProblemGetting an AI running on your own laptop takes four minutes. Everything that makes it actually usable afterwards is queueing, memory and patience — plain old engineering.
3Sep
Evaluation · 6 min readEvals Are the Only MoatAnyone can copy your prompts in a weekend. Nobody can copy the thing that tells you, precisely, whether your last change made the product better.
August 2026
2 essays27Aug
Retrieval · 6 min readYour RAG System Doesn't Have a Retrieval ProblemWhen an AI answers badly from your own documents, everyone blames the search. In every system I have debugged, the search was fine. The documents were the problem.
20Aug
Models · 6 min readI Fine-Tuned a 0.5B Model to Write Shell CommandsA tiny AI running on my laptop beats a giant one at a single job. That sounds like a story about size. It is really a story about picking a small enough job.
July 2026
1 essay10Jul
AI Interpretability · 14 min readInside Claude's J-Space: Anthropic's Global Workspace ResearchAnthropic found a small, editable workspace inside Claude that carries silent intermediate thoughts. Here is what the J-lens reveals, why it matters, and what it does not prove.
March 2026
1 essay27Mar
LLM Systems · 6 min readKV Cache from First PrinciplesA clear first-principles guide to KV cache: why it exists, how it works, and what tradeoffs matter in real production systems.