Building, breaking, and writing about the web.
For one week, I let internet strangers tear apart my portfolio and implemented their suggestions whether I liked them or not. Some changes made the site measurably better. One made it objectively worse. All of it taught me something about my own ego.
I tracked every dollar, every outage, and every migration scramble from two years of running josephkhan.me on free-tier everything. Spoiler: 'free' cost me about $400 and a very bad Tuesday in March.
I spent a week 'improving' my RSS reader instead of shipping the three features people actually asked for. Here's the hour-by-hour breakdown of how refactoring became my favorite form of avoidance — and the guardrails I built to stop it.
After three years of publishing on josephkhan.me, I finally exported my server logs and looked at the numbers honestly. The results humbled me, confused me, and taught me more about writing for the web than any advice thread ever did.
I handed my bio, case studies, and project blurbs to an AI for a full month and tracked what happened to my response rates. The results were messier — and more useful — than I expected.
I went through public job-hunt threads and found five portfolios with zero custom design that landed offers anyway. The common thread wasn't polish — it was shipping. Here's the case for deploying the rough version.
I spent years writing framework deep-dives that got ignored while a 400-word story about a botched deploy became my most-read piece ever. The data from my own site finally forced me to admit the stack was never the story.
josephkhan.me has gone through more rebuilds than a Detroit assembly line, and each version failed in its own special way. Here's the honest autopsy, version by version, and what actually changed once I started paying attention to how people read.
Four days into my first real dev job, I took down checkout for a company doing serious revenue. Here's what breaking production taught me about testing, owning your mistakes, and why failure stories beat success stories every time.
A URL shortener, a dead SaaS, a Discord bot, and seven others get scored on skills gained versus hours sunk. If you're choosing your next weekend build, this rubric might save you three months of your life.