Mobile + AI product engineer. 7 years of native Android (Kotlin/Compose) and iOS (Swift/SwiftUI/Combine), plus Flutter — and for the past year, shipping LLM products solo, end to end.
I work at the application layer of AI: RAG, context engineering, evaluation, and agent tooling.
📍 Seoul (UTC+9) · Open to remote · Full résumé →
passbaton · npm — Session-continuity hooks that give AI coding agents persistent memory across Claude Code, Codex CLI, and Gemini CLI, sharing one local SQLite db. Auto context injection, compaction handover, every feature opt-in. Listed in awesome-mcp-servers.
mac-pilot-mcp · npm — macOS UI-automation MCP server that learns and replays recipes.
Merged fix to coil-kt/coil (cache-strategy evaluation order).
Products live outside GitHub — a fuller write-up is in the résumé:
- Sajumung · Lunaday — two AI apps on Google Play, built on one structured-RAG + context-engineering engine I designed (abstracted across domains, not rebuilt).
- AI video pipeline — LLM script → keyframe → image-to-video → TTS → ffmpeg, with a consistency-metrics gate that rejects bad takes. Runs unattended. Sample.
- 4 web services on shared AWS infra.
- Solo, end to end — design → build → deploy → operate.
- Evals over vibes — LLM features get an LLM-as-Judge rubric, so changes are measured, not eyeballed.
Stack — Kotlin · Compose · Swift · SwiftUI · Flutter · TypeScript · Node.js · Next.js · Python · Docker · AWS