I'm all-in on TypeScript now. TypeScript gets first-class support for most modern web development tools and my interfaces are great AI food and guardrails. I love Python but don't use it much since I do 100% TypeScript webdev right now. But Python can be very good for backend processing tasks. I don't really write HTML directly anymore but I do throw the occassional HTML element into my React and Markdown pages. Shell I use daily for basic plumbing: installing packages, file operations, git operations, CLI tools, and all the shell scripts agents make and use.

This is my daily toolkit. React and TypeScript are what I'm writing most of the day, and I'm usually writing those inside of a Next.js app deployed to Vercel. Cloudflare is my registrar, DNS management, object storage, and CDN. GitHub Copilot is currently my AI coding assistant of choice (Claude Sonnet 4.5 for general work, Claude Opus 4.6 for tough problems, GPT-4o for natural language editing). VS Code is where I live. Chrome DevTools helps me with many aspects of my webapps. Jaygriff.com is pretty helpful too, highly recommend.

Turso is my primary database — SQLite on the edge with libSQL. Lightweight, fast, and perfect for my apps. Supabase I'm exploring for projects that need Postgres, auth, and realtime out of the box.

Tailwind CSS v4 is my styling foundation now, paired with CSS Modules for component-scoped styles. shadcn/ui (built on Radix primitives) is my component library — it gives me accessible, composable UI out of the box. The combo of Tailwind + Radix + shadcn is exactly the clean setup I was looking for.

I've also experimented with DaisyUI and Material UI in the past, and I used Emotion CSS-in-JS on the previous version of this site.

Notion is my main hub for writing and notes because it's fast for capturing lots of content and syncs across all devices. I admire Obsidian from afar (the hotkeys, settings, and aesthetic are amazing) but I don't actively use it. Surprisingly little use for spreadsheets these days given I'm a former accountant—most of my needs are met with Notion and custom coded tools. Current use case? Splitting expenses with my roommate using Google Sheets. The Microsoft Office Suite I have used for work when required. Locus is a custom Chrome extension I built for bookmark launching—I use it daily.

Affinity is my go-to for image and vector graphic work—professional-grade design tools without the Adobe subscription. I also use online SVG tools when I need quick edits. I do very little photo editing these days—when I do, it's practical stuff: cropping, aligning, background removal, optimization, and fixing aspect ratios and pixel sizes.

Recording demos of what I'm programming to display functionality, timelapse changes, and progress. OBS is powerful, free, and handles everything I need for capturing. DaVinci Resolve for editing when I need more than basic cuts.

I've used Figma for visualizing and wireframing sites. As a solo dev who's a programmer first, I found it less helpful than expected. I prefer vibecoding a rough version of the feature or page, then iterating on the design in code. It's hard to wireframe automated pages and systems and true interactions! Coolors helps me generate and explore color palettes, though I often just ask AI for color scheme suggestions.

Hammerspoon lets me automate macOS with Lua scripts—creating custom keyboard shortcuts and window management. AutoHotKey was for automating computer use on Windows: text expansion, window control, tool launching, and text manipulation. AI now does all the text expansion and manipulation I need.

These are the tools I'm actively watching in the AI coding space. The ones that made me realize AI isn't just a better search engine—it's an action engine that can do fine-grained work in the real world.

I'm excited about these tools but haven't carved out time to properly explore them. LangChain and LangSmith look powerful for building AI applications and agent workflows.

I don't run much personal infrastructure, but I study it extensively. I've realized that infra-aware software matters more to me than running infrastructure for its own sake. My own apps aren't at the level where they require or benefit from tons of personal infra—there are a LOT of ways to achieve things with hosted services these days. The main use-cases where homelab makes sense for me: GPU compute, performant mass storage, privacy, security, and learning.

My current plan for this site is to start implementing image and gif composition support throughout all my articles—I'll save the files locally, then host optimized versions in object storage. So a NAS is in my future. Another project I may need personal infra for (if a service doesn't meet the need) is sandboxed YOLO mode for AI coding assistants.