/uses page
My setup and tools
Everything I use to work — hardware, editors, apps, and services. Inspired by uses.tech.
🖥️ Hardware
- CPU
- AMD Ryzen 9 7950X3D — 16 cores, 32 threads. The 3D V-Cache makes a huge difference in compilations and frontend builds.
- RAM and storage
- 64 GB DDR5 + Samsung 990 Pro 2 TB NVMe. Never run out of memory with Docker, multiple projects, and the browser open simultaneously.
- GPU
- AMD Radeon RX 7900 XTX — useful for renders, image processing, and the occasional game.
- Monitor
- ASUS VG34VQL3A 34" ultrawide 1440p 180 Hz — the extra horizontal space is perfect for editor + terminal + browser without overlapping.
⌨️ Editor and Terminal
- Editor
- VS Code with extensions: Prettier, ESLint, GitLens, Tailwind CSS IntelliSense, PHP Intelephense. Theme: One Dark Pro.
- PHP / Laravel IDE
- PhpStorm for larger Laravel projects — the autocomplete and refactoring tooling is unmatched.
- Terminal
- Windows Terminal + PowerShell 7 with Oh My Posh. Git Bash for POSIX scripts when needed.
- Version control
- Git + GitHub. Lazygit for quick terminal operations.
🛠️ Main tech stack
- Frontend
- Astro 6, TypeScript, Tailwind CSS v4, Preact for interactive islands.
- Backend
- Laravel 12 (PHP 8.3+) as the main REST API. Node.js + Express for microservices and CLI tools.
- Database
- MySQL 8 in production. SQLite for local development and smaller projects.
- DevOps
- Cloudflare (DNS, CDN, Workers), Ubuntu VPS, Docker for reproducible dev environments.
📱 Apps and services
- Project management
- Notion for documentation and planning. GitHub Projects for technical sprint tasks.
- Design
- Figma for wireframes and UI design. Excalidraw for quick mockups.
- Task management
- Trello for Kanban boards with clients — simple, visual, and no learning curve for non-technical stakeholders.
- Communication
- Slack with larger clients and teams, WhatsApp for SMB projects. Zoom / Google Meet for calls.
- Cloud AI
- Claude (Anthropic) for code review and documentation. GitHub Copilot for in-editor autocomplete.
- Local LLMs
- The RX 7900 XTX has 24 GB VRAM — enough to run large models locally via Ollama with ROCm. I use Qwen 2.5 Coder and Llama 3.3 for tasks I don't want to send to the cloud: private client code, quick prompt testing, or just working offline.
- Analytics
- Google Analytics 4 for client projects. Umami (self-hosted) on my own site — privacy without giving up the data.