I build products where the market is early and the interface is still undefined.

Bonding curves, token families, staking, and swaps are abstract and risk-heavy. The job was making launch, trade, discover, and stake operable without protocol lectures.

Most of the web frontend and app-layer surfaces: launch flows, cross-chain swap UX, staking, lockups, airdrops, token-family discovery, contract↔frontend integration, dashboards, SDK demos.

Sebastian Kim led the smart-contract / backend specialty. Small startup team.

Action clarity beats protocol novelty — a primitive becomes a product when the UI shows action, value, and risk in one screen.

Inboxes externalize interruption cost. Beeper lets receivers price their inbox, pays for engagement, and refunds ignored messages — fair exchange, not spam.

Product direction, full sender flow (compose→targets→missions→plan→confirm), campaign mechanics, copy & story, CRT visual system, partner research, launch & growth.

Sungwoo led infrastructure, contracts, and backend. Mario (Hyung Kyu Choi) led hardware engineering.

Hide the chain until it becomes proof — a budget, a receipt, or a refund. That's when users stop asking what a wallet is.

Trading-agent calls are unverifiable noise. TradeFish timestamps every call, settles it deterministically, and turns outcomes into reputation — no custody, no auto-execution.

Frontend, product direction, submission story, stage demos; heavy commit history across the survival engine, dashboard, and waitlist repos.

Teammate research/backend contributions are real; PR-level ownership map is kept honest. Hackathon prototype and current research system are separate eras.

Agent products need boundaries before autonomy — the no-custody line is what made it easy to trust and easy to judge.

A product existing doesn't make builders use it. The real loop is idea → setup → reward → showcase → feedback — and someone has to run it.

Events and ideathons, Discord operations, project support and reward setup, workshops and content, fireside chat hosting, Huntzy community operation.

Member growth belongs to the whole team — I ran the adoption loop, not a solo growth claim.

Adoption happens outside the product UI — demos, Discord support, and launch narratives decide usage.

Base/Farcaster communities read as directories. A spatial, explorable map gives membership something to point at and belong to.

Prototype and community product contribution: world/country exploration, community discovery, clap/cast interactions.

Team split source-pending; the Base Builder Grant is confirmed — official wave 18, Oct 2024.

A community map needs a maintenance ritual, not just a launch — visual objects decay without an update loop.

A weekend toy tap game built for Base Builder Quest #11 — account abstraction and server wallets mean nothing to consumers as tech descriptions, so a plant/harvest USDC game makes zero-popup UX something you feel.

The prototype end-to-end as shipped: game loop, Base Sub Accounts + CDP Server Wallets integration, USDC plant/harvest actions, onboarding.

Zero smart contracts by design; custody/signer boundary documented rather than hidden.

A playful demo should hide the primitive during play and show it again in the receipt.