AI consulting for Korean companies entering the US.
One consultant, both languages, both markets. Based in Los Angeles, working remotely with teams in the US and Korea.
I work in English and Korean, remotely, with clients in the US and Korea. A lot of my work sits on that line: teams in Seoul building for the US, US companies working with Korea, founders who think in one language and sell in the other. Nothing gets lost in translation because there isn't any. You get the same person on both sides.
I build autonomous AI systems that run companies end to end. Not demos and not chatbots: systems of agents that plan, execute, check their own work, and escalate only the decisions a human should make.
- Korean companies entering the US
- Your product, docs, and customers are splitting across two languages, and the US side keeps getting a translated summary of the Korean original. I build the AI systems and workflows for the US operation directly, in English, and report back to Seoul in Korean. Support, internal operations, agent systems: built once, correct in both languages.
- US companies working with Korea
- You have a Korean team, partner, or parent company, and the work slows down every time it crosses the language line. I sit on that line. Requirements, specs, and systems get built with both sides in the room, because I am both sides in the room.
- AI System AuditFrom $1,500
- A full read of your AI system as it actually runs: architecture, agents, evals, guardrails, and the places where context gets lost. You get the findings and a fix plan you can hand to any engineer, including the one who built it.
- Agent Architecture ReviewFrom $1,500
- A design review of an agent system before you build it, or a second opinion on one that is already designed. Orchestration, memory, tool boundaries, and how the system proves its own work, checked against what actually breaks in production.
- Autonomous Workflow BuildFrom $2,500
- One workflow, built end to end: an agent system that takes a recurring operation off your team's plate, with the checks and escalation paths that let it run unattended. Scoped together first, then built with receipts.
Fixed scope, fixed price, async by default. No hourly billing, no staff augmentation.
- Fractional CTO at an AI-first education company, where an autonomous system I built supports operations end to end
- KERNEL, my open-source plugin for coding agents: active on the Claude plugin marketplace, with real external users
- Three free apps live on the App Store: ModelMind, Paper Rooms, and Civic Forges
- Open-source evaluation and memory tools, published on GitHub and PyPI
Send the messy version. I will look at the project, the architecture, and what to build next, and reply by email. Free, one time, async.