AI Platform Engineer - Agent Harness & Web Automation

KWAVE is building the AI-native operating system for K-commerce in the US: a B2B platform where Korean brands, US retail buyers, and individual sellers run their stores with personal AI agent teams. Every workspace gets a staffed team of AI agents that handle sourcing, marketing, operations, and influencer campaigns - connected to K-Live, social platforms, logistics, and payments., You will own two things:

• Site & content auto-generation. Our marketing surfaces (kwave.ai and the per-vertical pages, including the influencer-marketing product) are evolving from hand-built pages into an agent-driven pipeline: agents that generate, localize (en/ko + 4 more), and ship landing pages, campaign pages, and seller storefronts. You will design and build that pipeline.
• Codebase structure & internal quality. Our monorepo has grown fast. You will lead refactoring, module boundaries, conventions, and the docs/automation that keep an AI-assisted codebase coherent - so that both humans and agents can navigate and extend it safely.

You will work directly with the founder and with Claude Code agents daily. A meaningful part of the job is making the agents better: tuning prompts and instructions, building the harness (tools, guardrails, evals, CI gates) that lets agents do more of the routine work reliably.

What you’ll do

• Build the homepage/landing auto-generation pipeline (Next.js App Router, i18n dictionaries, design tokens) from structured product data
• Refactor and re-structure the TypeScript monorepo; define module boundaries, lint/type gates, and contribution conventions
• Develop and maintain the agent harness: MCP tools, prompt/instruction templates, evaluation suites, and regression gates for agent output
• Integrate external APIs behind our connector layer (social platforms, creator-data APIs, video-generation APIs)
• Own quality: typecheck/CI discipline, i18n parity, visual regression on marketing surfaces

Do you have experience in TypeScript?, Do you have a High school diploma or GED?, * 4+ years of production software engineering, with strong TypeScript and React/Next.js

• Experience designing or significantly restructuring a real codebase (monorepo, module boundaries, migrations) -not just feature work
• Hands-on experience building with LLM APIs: prompt engineering, tool/function calling, structured output
• Daily fluency with AI coding agents (Claude Code, Cursor, or similar) - you can direct, review, and correct agent output at speed
• Self-directed; comfortable owning an area end-to-end in a small team

Nice-to-have

• Agent harness / agent framework development: MCP servers, multi-agent orchestration, eval pipelines, instruction tuning (what we mean by “AI tuning”)
• Fine-tuning or systematic prompt-optimization experience
• tRPC, Drizzle/Postgres, Tailwind, pnpm monorepos; Vercel/Railway/Fly.io deployment
• i18n at scale (we ship en/ko first, 6 locales total)
• Korean language ability or familiarity with K-content/commerce (helpful, not required)

How we work

• AI-native: agents write a large share of the code; engineers set direction, review, and build the rails
• Small team, direct communication, fast iteration - work ships to production daily
• English or Korean working language; both is a plus, * High school or equivalent (Preferred)

Pulled from the full job description

• Flexible schedule, * Flexible schedule