Experiments, side projects, and notes from the dev lab.
I rebuilt my entire side project toolchain under the constraints of a 1968 mainframe: 1MB of RAM, no cloud, no autocomplete, no mercy. Three weeks later, my dev habits look very different — and honestly, so does my code.
For one week, anyone could push to my side project. Anonymous contributors renamed functions, restructured modules, and left commits like 'fix' and 'trust me.' Here's the full crime scene report, including the changes I'm actually keeping.
I wired an LLM into my homelab's monitoring stack so every service could file its own outage report in plain English. It worked great — until the reports got mouthy and I had to decide how much personality infrastructure is allowed to have.
I wired a Raspberry Pi and a camera module into my old intercom, and somewhere along the way the project stopped being about hardware. Once my neighbors got access to the doorbell feed, the real engineering problem turned out to be people, not solder.
For 48 hours I let strangers hammer, poke, and outright attack my live side project — and kept a running log of everything that broke. Spoiler: my security assumptions died faster than my uptime.
I wired a Raspberry Pi to text me whenever something in my home lab breaks, and then I gave the notifications a personality. A month of passive-aggressive alerts later, I know exactly which failures actually deserve my attention.
Ten years of commits, branches, and abandoned repos turned out to be the most honest productivity journal I never kept. Here's how I ran the analysis, what the 2am commits had in common, and the three habits that came out of it.
Half a year of daily engineering notes, dumped into one file and actually re-read. What I found: ideas that took three tries to stick, dead ends that were just early, and one debugging blind spot that cost me entire weekends. Plus the dead-simple note template that made the audit possible.
I released a tiny CLI tool as open source and committed to treating it like a digital pet — daily feedings, weekly grooming, full emotional investment. Here's the honest ledger of hours, stars, and what it actually cost me.
Perfumers can tell a good fragrance from a cheap knockoff in one whiff. Developers should be able to do the same with frameworks. Here's the five-sniff test I run before anything gets a repo in my lab.
Hosting, domains, API credits, a Raspberry Pi that never sleeps — the real operating cost of a one-person dev lab, itemized. Plus the ROI rule I use to decide which experiments earn compute time and which get a graceful shutdown.
I dragged five half-dead experiments out of my repos folder, gave each one a README, a demo, and a two-page write-up, and stapled them to a single landing page. Three weeks later, two interviewers brought up the parking app unprompted. Here's exactly what I changed and what recruiters actually clicked.
We celebrate launches and quietly ignore the projects we abandon, which is exactly backwards. Killing a side project — deliberately, with a post-mortem — is a judgment skill worth putting on display. Here are six funerals from my own graveyard and what each one taught me.
I ran my own homelab for twelve months and kept receipts on everything — power, parts, drive failures, and the hours I'll never get back. Here's the unvarnished math, what the same budget buys in the cloud, and how to tell if your server rack is a hobby or a habit.
A working list of seven experiments I'm running this year, from local LLMs on a mini PC to a $15 garage automation hack. Each entry has a hypothesis, a tool stack, and a difficulty rating — steal one and build along.
Every developer has a folder full of half-baked experiments nobody asked for. Mine includes a Slack bot that graded tacos and a CLI that guessed when my laundry would finish. Here's why I'm done apologizing for them — and why you shouldn't apologize for yours either.
Most developer portfolios are digital business cards gathering dust — built once, never measured, never improved. I rebuilt aleksdelab.com like it was a product with users, and the before/after numbers convinced me this is how every dev should think about their personal site.
I challenged myself to ship one tiny project every week for a month — no polish passes, no rewrites, no excuses. Here's the honest log: what launched, what flopped, and why shrinking the scope on day three saved every single release.