AI Engineer

Prompt Architect tracks and optimizes brand visibility (Share of Voice) inside AI answer engines like ChatGPT, Perplexity, and Gemini. As an AI Engineer you'll move across pa-web (Next.js 16), pa-server (Fastify), and packages/db (Drizzle/Postgres) — using Claude Code and our internal agent stack as your primary input devices — and ship features end-to-end, from idea to production.

We want people who learn tools fast and design the workflow itself. The reason we ship in a week what other teams ship in a quarter is not luck. It's the system.

• Ship full-stack features across frontend, backend, data, and infra — Next.js App Router, Fastify, Postgres, pgvector.
• Design and operate GEO engine pipelines — prompt execution jobs, mention matching, entity extraction, RAG / knowledge graph.
• Agent orchestration — Claude / OpenAI tool-use, durable workflows, evals, cost & latency monitoring.
• Treat Claude Code, Cursor, and our MCPs as production tools — your skills, hooks, and automations lift the whole team's velocity.
• Own observability for what you ship — Sentry, Axiom, structured logs. You instrument it, you read it.
• Run the spec → architect → build → QA pipeline yourself (with subagents).

• 3+ years shipping production software — frontend and backend, on your own hands, no silos.
• AI tools deeply wired into your daily workflow. You can articulate why one prompt outperforms another.
• TypeScript, Node.js, Postgres, and React at native fluency.
• Comfort with ambiguity — you can take 'build the careers page' and ship in a week without 200 questions.
• Production-grade from day one. No `stub-id` placeholders, no hardcoded fallbacks.
• Clear written PRs / specs in English or Korean (both is better).

• Built MCP servers, Claude Skills, Cursor rules, or custom subagents.
• Production experience with RAG, vector DBs, or embedding pipelines.
• 0 → 1 or 1 → 10 startup experience.
• Background in dev tools, data tooling, or growth infrastructure.

1. 0130-min introCasual chat with a founder. We'll skim your resume, portfolio, and GitHub.
2. 02AI workflow demoWalk us through something you built recently. How you used AI is the point.
3. 03Paid pair build (4h)Real work in our codebase, together. We pay for your time.
4. 04OfferWe decide within a week.

• Cash + meaningful equity at early-stage scale.
• MacBook M4 Max, dual 4K monitors, Claude Code Pro, Cursor, Linear, GitHub Copilot — all on us.
• San Francisco, London, and Seoul offices — hybrid or fully remote after onboarding.
• Unlimited budget for AI / engineering books, conferences, and seminars.