Engineering & Infrastructure
TypeScript, Python, Docker, Kubernetes, CI/CD, E2E
Ten years of full-stack before any of it was called AI. It is still the part that decides whether the clever bit ever reaches a single user.
- 3,300+commits in twelve months, across 33 repositories
- 10 → 10production incidents: a day, then a month
- −30%hotfixes, after automated E2E entered the pipeline
Full-stack, in the boring literal sense
TypeScript, React, Next.js, Node.js, Python, PHP with Symfony and Laravel, C#. Ten years of it, and still shipping daily — the management titles never came with a keyboard being taken away.
Deployed the same way twice
Docker, Kubernetes, GitHub Actions, nginx with automatic TLS, monitoring. On-premise and fully self-hosted when that is the requirement rather than the fallback. My first job had no deployment tooling at all, so I built its CI/CD pipeline and a multi-instance release manager; I have not shipped without one since.
Tests that gate the release
Unit and E2E suites with Playwright and Cypress, in the pipeline rather than in a wiki page. Introducing them on a B2B ordering platform cut hotfixes by 30%, and the QA analyst I trained to run them became the company’s lead QA.
Making a live system boring
One B2B SaaS went from ten production incidents a day to ten a month. There was no rewrite. It was instrumentation, typed contracts at the seams, and refusing to ship the fix before understanding the cause.
Stack
- TypeScript
- React
- Next.js
- Node.js
- Python
- Symfony · Laravel
- C#
- REST
- Docker
- Kubernetes
- GitHub Actions
- Playwright
- Cypress
- nginx · TLS