Three years of writing code, shipping products, and occasionally taking things apart just to understand how they work. Lately: AI, education technology and developer tooling.
Software isn't writing code to me; it's chasing a problem.
Designing how it gets solved, building the system, deploying it, and watching people actually use it — that is the best part of the job. Rather than specialising in a single layer, I like being everywhere from the idea to the infrastructure. The fastest way to understand something is to build it.
An AI-native education platform. It rethinks how students, teachers, coaches and institutions work with AI from the ground up: assistant, homework and quiz flows, teacher workflows, analytics, gamification, institution management and model routing.
The terminal, rethought for developers. Split panes, AI workflows, file management, system information and tooling in a single workspace. Open-source at the core, ecosystem on top.
An AI-assisted audit CLI that watches a project continuously, catches likely security issues, reports them and proposes fixes. An attempt at making security part of the pipeline instead of a separate job.
An AI-assisted editor for resumes that make it through ATS. The user only writes; the system handles formatting, keywords and structure.
Not everything has to become a product. Some of these were only ever answering a question.
Spent writing code by hand and delivering freelance work. I have always learned through projects rather than around them.
Of restaurants run on a QR menu system I built. That is where I first learned what “real users” actually means.
From interface to server. Linux, Docker, CI/CD — I want to understand the place a product actually runs.
Not closing tickets, but deciding where the product goes. Mentora is the largest attempt at that so far.
Most of the tools I rely on were written by someone else for free. Paying that back runs through the same place.
See on GitHub →Product, software, AI, or just a good idea. We can talk — you do not need a project budget to get a reply.
I read every message and reply personally, usually within two working days.
Your details are emailed to me and never stored in a database, shared, or added to a mailing list.