Designing systems.
Shipping products.

The problem. Sam Spratt's LUCI universe spans four collections and 929 tokens — ten 1/1 paintings, fifty gifted skulls, 256 Monument Game players, 613 masks — but its ownership stories lived scattered across contracts and marketplaces, invisible as a whole.

Design decisions. After a phone call with Sam about overhauling his site, I shipped this explorer as my vision for the redesign: the ecosystem as a heliocentric system — the 1/1s at the center, each layer orbiting outward in the order the world grew. Sam adopted the heliocentric model as a central part of his new design. The map is navigable — hover any token and its collector's network lights up; click through to a collector dossier — and the rings were engineered to render live on-chain data via Alchemy's NFT API, a feed I've since retired to a snapshot to keep an unofficial community tool cheap to run.

Shipped result. A community tool for exploring a $12M+ art ecosystem — 929 tokens, 81 cross-layer collectors — that reads like a map of belonging rather than a block explorer.

The problem. Good Vibes Club — a 1,600-holder NFT community from animation studio Toast — needed one honest, always-current picture of its market: sales, holders, liquidity, whales. Nobody should have to compile a report.

Design decisions. A numbered editorial system (01 Price & Volume through 04 Collection Health) turns a dashboard into a publication regulars learn to read. Every visualization has an export button — and it doesn't hand you a screenshot, it renders a socially optimized graphic: branded, annotated, and sized for the feed, so any chart or sale is one tap from being shared. ETH/USD and time-range toggles persist across the page. The site refreshes itself and says so — "Updated 18m ago" is a design element, because trust in data is trust in timestamps.

Shipped result. 13,878 ETH (~$26.8M) of market history live at a glance.

The problem. A top-5 decentralized exchange — $10B in transaction volume, 300,000+ users onboarded — was limited on analyzing growth due to scattered ad-hoc queries.

Design decisions. Executive views styled like finished presentation slides but wired straight to the database — one click refreshes the whole deck. Day-over-day deltas are annotated on every bar so trend and rate arrive together. Users are bucketed by XP earned, exposing the silent majority a topline average hides.

Shipped result. Weekly hours of manual deck-building, gone — and the XP-bucket view surfaced most users transacting near zero, the analysis behind my recommendation for threshold-based incentive rewards.

The problem. Artist m0dest's on-chain elimination game — 6,969 citizens paying taxes, auditing rivals, buying insurance, surviving — had all of its drama locked inside raw transactions.

Design decisions. A bureaucratic terminal: phosphor green on black, tabulated activity logs, a ticking treasury counter. The interface itself delivers the game's deadpan joke — nothing is certain except death and taxes. Kill leaderboards come from a subgraph, every state is verified on-chain, and the room refreshes every 30 seconds so it always feels inhabited.

Shipped result. The community's shared window on the game — day 143, 99.5% eliminated, 69 survivors, an 816 ETH treasury — designed to be watched like a scoreboard.

The problem. Meal planning was a thirty-minute Sunday negotiation that ended in the same five boring dinners.

Design decisions. The whole product is one promise — plan your family's meals in 60 seconds — and every screen supports it. Pick dinners from 225+ curated recipes (or import your own from any site), pair sides that quietly track your plate's color variety, then one tap produces an organized grocery list. Friendly, rounded, zero-jargon — designed for the least technical person in the household.

Shipped result. Live and free — no credit card, no subscription — and wired to an autonomous agent (OpenClaw) that can turn the week's list into a placed Instacart order.

The problem. NFL research expires weekly. Publishing it by hand — new charts, new layouts, every week — doesn't scale past a hobby.

Design decisions. One visual grammar, expressed across every data shape: team-colored bars, a constant header band, the boot mark in the corner. Consensus-ranking boards, single-game reviews with cumulative EPA and pass-dispersion grids, league-wide tendency quadrants — each is a template that renders publication-ready straight from the data.

Shipped result. An automated research platform that turns each NFL week into finished graphics. Same system, any story — the design-system instinct applied to sports data.