XVII The Tinkerer
Software engineer · tinkerer · researcher
Curious about what happens underneath.
I build software, take it apart, and try to work out why it held together in the first place.
I'm a software engineer interested in the space between building things and understanding them. My work runs through web development, backend systems, infrastructure, AI and security — usually because something wouldn't stop bothering me.
Outside of code I'm drawn to things with character: old cars, photography, mechanical objects, and the occasional card trick.
Different subjects, same question. How does it work?
Method
Attributes
Everything I work with falls into one of four piles. The pile matters more than the logo.
Systems & infrastructure
Containers, reverse proxies, queues, CI, the machine underneath the machine.
AI & language models
Building software around models, and asking hard questions about what they produce.
Security
Authentication, static analysis, vulnerability detection. How systems fail, and why.
Web & craft
The part people actually touch. Interfaces that behave like well-made objects.
I needed something. So I built it. The interesting part was usually somewhere in the middle.
I needed a CLI TOTP client so I built one.
A secure, cross-platform TOTP manager built in Go for generating and managing 2FA codes directly from the command line. Features encrypted secret storage using AES-256-GCM, password-based protection, and both CLI and interactive modes.
Why n8n only supports telegram bots?
An n8n-powered Telegram automation bot that connects Telegram interactions with APIs and automated workflows. Designed as a modular, event-driven system for streamlining tasks and integrating external services.
If a language model can write software, can it also help secure the software it writes?
A framework that puts security analysis between the model and the developer: generated code is passed through static analysers, and the findings are fed back to the model so the next revision is measurably safer than the first.
The practical question underneath it: as more code is written by models, who is reading it?
Notes from things I couldn't stop wondering about.
Not a comparison table. What a header actually carries, what a cookie actually is, and where the token really lives.
Objects that can tell you when something happened, without you having to ask them repeatedly.
One repository, two runtimes, one reverse proxy, and the small decisions that decide whether the next deploy is boring.
Cars. Cameras. Machines. Things with interesting mechanisms.
Nothing developed yet. Come back after the next roll.
Working on something strange, technical, or both? I read everything.
Write to meGo on. It's not a real trick.