MY SETUP

Tools I use every day

Curated gear and software — rough notes on why I chose each one. Updated monthly.

Updated
[object Object]
HARDWARE

Computers & peripherals

The machines I tap on every day and the gear that makes them bearable.

💻

MacBook Pro M4 Pro · 48 GB

Primary workstation. 48 GB RAM keeps the full monorepo dev stack (Vite, Storybook, Turbo cache) plus local Ollama models running in parallel — no swap, no slowdowns.

🖥️

Samsung Odyssey Neo G9 · 57″

7680×2160 super-ultrawide — effectively two 4K displays on one curved canvas. Kills the need for dual-monitor alignment.

🏋️

Cougar E-GRAV 160/180

Electric sit-stand desk with dual motors, carbon-fiber laminate, full-desk mousepad, and cable tray. Two memory presets for sit/stand switches.

🪑

Razer Iskur

Gaming chair with integrated lumbar support. Survives 10-hour debugging marathons without wrecking my back.

🧘

Yoga ball

Rotation partner for the fancy chair. Forces micro-movements and core engagement during long call blocks. Cheap posture insurance.

EDITOR & TERMINAL

Where I write code

Short list — tooling should get out of the way, not become the work.

🤖

Claude Desktop + Claude Code

Anthropic's GUI + CLI. Opus 4.7 with its 1M-context window powers this repo's agent harness (see .claude/).

🌌

Google Antigravity

Google's multi-agent IDE. Parallel agent orchestration for exploratory work and research sweeps.

⚡

Zed

Rust-based editor with multiplayer editing and first-class LLM integration. Absurdly fast — feels like pressing keys directly onto the screen.

SHELL & CLI

Terminal companions

🐚

zsh + starship

Fast prompt with git, node, and pnpm indicators. No Oh-My-Zsh — starship renders faster and its config is a single TOML file.

🐙

gh

GitHub CLI. PRs, releases, workflow runs, secrets, and issue triage without leaving the terminal.

☁️

gcloud

Google Cloud CLI — Cloud Run deploys, IAM changes, log tailing, and WIF provider debugging for this project's infra.

📦

pnpm + Bun 1.3

pnpm for the monorepo (workspace: protocol, frozen lockfiles). Bun 1.3 for one-off scripts and quick test harnesses — 30× faster installs, zero-config TypeScript.

🔍

ripgrep + fzf

`rg` for code search (respects .gitignore, fast enough to be instant). `fzf` for interactive file and history picking — piped into almost everything.

AI & LLMS

Where intelligence lives

A mix of hosted frontier models and local runners. The goal is to know each tool's edge well enough to pick the right one fast.

🎬

Google Flow

Google's AI video generation surface. Scene-by-scene storyboarding with camera control — my go-to for social clips and post hero animations.

🍌

Nano Banana

Google's image generation model. Hero images, blog illustrations, social graphics — created and iterated conversationally via text prompts.

🦙

Ollama

Local LLM runner. I use it to benchmark compact open-weight models — tracking when they become capable enough to power specific personal projects without relying on hosted APIs.