Founder stories · 3 min read · 11 October 2026
From web design to AI: eleven years of switching stacks
HTML and Photoshop in 2015, Laravel, React, testing, automation, Python and now AI. What each stage taught me that I still use.
I started in 2015 designing websites. Today most of my time goes into Python, AI and the products I run under XavierStack. In between, I've switched stacks more times than I planned to. Each switch felt like starting over. Looking back, every one of them left me with something I still use every day.
2015 to 2017: web design
HTML, CSS, JavaScript, Photoshop, and later Figma. I designed landing pages and portfolio templates, and redesigned an e-commerce site.
What stuck: users don't read, they scan. A page lives or dies in the first screen. I still sketch every product screen before I write a line of code.
2017 to 2019: PHP and Laravel
This is where I moved to the backend. I built more than 15 Laravel applications, custom CMS work and payment gateway integrations, and properly learned MVC.
What stuck: structure beats cleverness. Laravel taught me that a predictable folder layout and clear models matter more than any trick in the code.
2019 to 2020: React and modern frontend
React, Next.js, Redux, TypeScript and Tailwind. Dashboards, admin panels with complicated state, and single-page apps.
What stuck: state is where bugs live. The less of it you keep, the fewer surprises you get.
2020 to 2021: automation testing
Cypress, Selenium, Playwright, Jest and Puppeteer. End-to-end suites, tests wired into CI/CD, and performance testing.
What stuck: if it isn't tested, it's broken, you just haven't found out yet. It's also where I learned to love CI pipelines.
2021 to 2022: workflow automation
n8n, Make, Zapier, IFTTT and webhooks. CRM integrations, data sync between tools, and alert systems for businesses.
What stuck: most business problems aren't code problems. Half the time, connecting two tools that already exist beats building a third one.
2022 to 2024: Python
Django, Flask, FastAPI and Pandas. Data pipelines, REST APIs and web scraping. Django became my default backend.
What stuck: "batteries included" is a feature. Django's admin alone has saved me weeks on every marketplace I've built.
2024 to 2026: AI and LLMs
RAG systems, prompt engineering, LangChain, LlamaIndex, vector databases and Hugging Face. A production RAG chatbot, multi-agent setups, and a lot of work on cutting token usage.
What stuck: AI is only as good as the data and the guardrails around it. The model is the easy part.
2026: building my own products
Now I'm founder of XavierStack, building the web versions of LocalUncle, BookMeDrive and mydockspace. Every stage above shows up in that work. Design decides whether people trust the product. Laravel and Django habits keep the code sane. Testing and CI keep me from breaking things at midnight. Automation handles the busywork. AI is starting to show up inside the products themselves.
No stack is wasted. You're not starting over, you're stacking.