an open notebook · v.2026
Brain Drip.
An open notebook on modern machine learning — written by one person, in long-form, with no schedule.
01Four kinds of writingthe library
→
56
interactive explainersDrips
One concept at a time. Read these when you want to understand a single idea well — each is self-contained.
Most recentNatural Language Autoencoders
→16
read in orderCourses
Sequences of Drips arranged in a teaching order. Read these when you want to learn an area from the beginning.
Most recentLLM Concepts
→06
essays & argumentsNotes
Essays, observations, and arguments. Read these to know what I think, not just what I know.
Most recentFour Agent Blueprints You Can Run in Under 80 Lines
→31
working codeBlueprints
Step-by-step build guides. Read these when you want working code by the end of the afternoon.
Most recentRun an Open Model with Ollama
02Recent entrieslatest first
01Jun 2, 2026NoteFour Agent Blueprints You Can Run in Under 80 Lines9 min02May 20, 2026NoteAI Agents Work. The Data Doesn't.03May 14, 2026DripNatural Language Autoencoders9 min04May 14, 2026NoteState of OSS AI, Spring 202605May 13, 2026NoteThe /goal command — 11 days, 3 tools, one primitive06May 9, 2026NoteWhy I'm redesigning Brain Drip6 min07May 9, 2026NoteAirbnb's 60% playbook08Apr 22, 2026DripTransformers, From First Principles14 min
why it exists
A small collection of explanations I’d have wanted to read when I was learning these things.
AuthorOne person
ScheduleNone
AssumesYou can read code
PriceFree, always