Writing
Notes on coding agents, the tools I build, and backends.
- The bill for free softwareNewAI agents made writing code nearly free, and I'm all in. But building with them, and building products that run on AI, takes more expertise, not less. Free software is real for an app with one or two users. An app with thousands needs a different approach — and it has a price.
- Coding isn't dead. It's where the loop closed first.New"LLMs write code now" is a statement about a feedback loop, not about a model. Why coding was the first domain to fall, why the same approach reaches everywhere, why typing is dead and programming isn't — and what that leaves for people who came here because technology is interesting.
- A harness for an AI-native app: evals, labs and a fast judgeNewWhat changed when every decision a model makes in my app became measurable — what evals actually test, how labs turn hunches into hypotheses, and why a fast, cheap decision model turned out to be the breakthrough.
- Coding agents: the boost is real, and so is the plateauNewWhat a year of working with coding agents gave me, where they stop helping, the traps worth keeping in mind — and the one habit that changed how I work with them.
- Shipping a dev tool as one binaryNewWhat I learned porting procdeck from Node to Go — embedding the web UI, releasing by git tag, and landing in npm, Homebrew, and Linux package managers without a desktop app or a signing ceremony.
- procdeck: debug from observability, not from codeNewI built a process manager so my dev stack — and my coding agent — could answer "what just happened?" from the runtime instead of re-reading the code.
- How the JavaScript event loop worksNewCall stack, microtasks, macrotasks, the render step, and Node's libuv phases — with an interactive step-through visualizer. The interview answer, from first principles.
- A chat is an agent with a human inside the loopNewTrying to find a place for AI agents in my life, I realized the difference between a chat and an agent isn't the model — it's the harness around it.
Job-search guide
Resumes, ATS, cover letters, tailoring and a search strategy.
- Building a simpler AWS deploy toolNewWhy I snapped and built my own AWS deploy tool — one that skips CloudFormation entirely and ships resources with direct API calls.
- Best use of LLMs isn't codeNewI'm not sold on the LLM hype — but they changed one thing for me, and it has nothing to do with writing code.
- Hiring without feedback is brokenNewPass the screening, answer every question, get rejected with no reason. You can't debug a process that returns no error message.
- Automate your resume, don't write itNewA resume is a build artifact, not source code. Why I stopped rearranging bullet points and wrote the source for my career instead.
- From 85% rejections to interviewsNewMost of my applications got auto-rejected. The fix wasn't a better resume — it was a better process for reading job listings.
- The generalist's hiring problemNewI can pick up any stack — but keyword filters are keyword-shaped, and people aren't. Why hiring rewards narrow specialists and punishes curiosity.