A Frustrating AI Code Smell: Every Problem Decomposes to a dataclass With Four Reasons
Turning messy evidence into a clean taxonomy before it has earned one.
Hallucination Station
Autoregressive thoughts on ML, AI, and shipping things. By Alex Furrier.
Latest posts
Completely Impractical but Delightful: An E-Reader Attached to My Phone
Using AI to turn a $70 accessory into a literal attention shield.
If AI Coding Trends Continue, I’ll Be Merging 20 PRs a Day
I went from merging 3.4 personal PRs a week to 19.2. Extrapolate that another year and the math says 20 PRs a day. Yeah, right.
I Put My Dog in My Coding Agent
Pi made it easy to put a sprite version of my dog in the terminal, then bundle the small coding-agent features I wanted around it.
Dogfooding Is the New Code Review
AI agents can ship code faster than humans can review it. Correct code still needs human time to become good software.
An Update on Codex's Gremlins
OpenAI explained why Codex kept talking about goblins. Of course it was reward hacking.
There are Gremlins lurking in Codex GPT-5.5
Codex GPT-5.5 apparently needed an explicit, repeated reminder not to talk about certain creatures unless they are actually relevant.
The Shiny Object Trap
When you can do anything, what you don't do is the skill
Karpathy's LLM wiki looked familiar. Here's how to bootstrap your own in 5 minutes.
Andrej Karpathy described exactly what I've been doing for the past few months. Here's a template to get started in five minutes.
Rummaging through Claude Code's spring cleaning closet
Types as agent instructions, auto-generated verifiers, and experimental REPL mode.
It shouldn't take 50,000 buttons to format a bulleted list
Software UX has been built around make it easy to do things. It's shifting towards make it easy to say what you want.
It takes 7,500 tests, 12 languages, and 3 browsers to make text wrap around a circle
One person and one AI built pretext in 27 days. Autoresearch applied to a real problem.
Agent-Friendly CLI Design
10 principles for building CLIs that AI agents can actually use, from a Cursor dev. Plus the missing piece: WHEN and WHEN NOT to use each command.
LLMs Are Cowboy Coding
Anthropic's harness engineering article highlights what's missing from most AI coding: scrutiny and evaluation.
Don't RAG Until You Have To
Why build a RAG pipeline when AI can already do most of the work for you?
ML Monitoring: Keep It Simple, Keep It Compounding
You don't need drift detection to start. One dashboard, one runbook, and a process that gets a little better every time.
Ruining the Fun of March Madness by Having AI Fill Out My Bracket
I pointed Claude at Evan Miya probabilities and Yahoo public picks and told it to build a bracket optimizer. It mostly worked, except for the part where it started picking 16-seeds.
Brandolini's Law Is Flipping
AI makes checking claims cheaper than making them. A penny-rounding simulation proves the point.
Don't Replace Your Job, Replace Your Chores
Automate the draining parts, focus on what gives you energy
"The CLI is Dead"
Agents + CLIs + skills is the core of the future - nobody has figured out the right human interface for that yet
My AI Dev Flow in 10 Bullet Points
No theory, just the setup — how I actually work with AI tools day to day
Agents Only Need Four Tools
Everything else is CLIs and skills files
Notes on Self-Hosting LLMs (Early 2026)
What to expect when running local models on consumer hardware
Muon: The Magic of Three Numbers Making GPUs Go Brrr
The Muon optimization algorithm stablizes multi-million dollar GPu training run and does so on the strength of three "magic" numbers
How Claude Code actually uses CLAUDE.md files
Discovered that CLAUDE.md files get passed to Claude with a disclaimer that they might not be relevant - explains why project instructions sometimes get ignored
Homelab Update: The Paternity Leave Edition
Five years of accumulated homelab jank finally gets some love during paternity leave
Firecrawl: A Simple Web Scraping Script
A command-line utility for scraping web content through the Firecrawl API with automatic file naming and content cleaning