Founding Platform Engineer — Arcade

• Location: Remote
• Sector: Blockchain
• Source: web3.career

About Arcade

We’re the team behind PRJX, the #1 DEX on HyperEVM, which generated $5.4M revenue in year one with 2 full time people.

Now, we’re pursuing the biggest opportunity in crypto.

The foundation of which is creating the best place to trade crypto (spot + perps) on your phone.

On top of that, we’re building a completely original social product with several novel mechanisms.

We’ll reveal more to promising candidates, but whatever you think it may be, think bigger.

Role

Staff+ level · $200,000+ · significant equity

Current stack. TypeScript/Node · Express 5 · Postgres with Drizzle · pnpm monorepo · Awilix · Vitest · AWS (SQS, ECS, Lambda, RDS, S3, CloudWatch) · Solana via @solana/kit · viem. You’ll have the authority to rewrite what doesn’t meet the bar.

You’re the first and only engineer in your domain. Your job is to take an original product vision and turn it into a robust system, holding the entire architecture and smallest details at once.

This is a role for top 1% devs. You’d trust yourself to architect this from zero to $100M/day in two months, and fix critical prod bugs at that scale the hour they land.

You will…

Own the platform: data layer, rewards, notifications, the API surface, and its reliability.

Integrate the trading engine: perps through Hyperliquid, spot through Jupiter and Relay.

Turn providers that disagree (Zerion, CoinGecko, Mobula, Alchemy, Hyperliquid) into one source of truth.

Serve heavy data fast: caching that survives a stampede, fan-out that holds.

Make one USDC balance across chains invisible to the user.

Hold the line at peak: apps in this category lag and crash; ours can’t, and a user can always close a position.

Kept a system alive through a 50x spike in ten minutes, and decided in advance what to shed.

Built rewards and accounting that survive retries: idempotency, replay safety, lot tracking.

Accounted for a latency budget end to end: tap to on screen.

Fixed Postgres bottlenecks in Postgres, not with a second database.

Deep in a typed backend language (TypeScript, Go, or Rust), and productive in a codebase that isn’t yours; EVM at read level.