I build AI-powered tools and ship fast. Currently creating FlowCraft — a multi-agent AI platform that builds Flutter apps from natural language.
Participating in the Global PBL Program — a 6-month intensive in full-stack development, applied AI, and startup building. Competitively selected. Shipping projects daily.
Kookmin University, Seoul — expected graduation 2026.
A platform that lets anyone create and run Flutter apps from a single English sentence. I designed a 6-agent AI pipeline (Orchestrator, Planning, UI, Logic, Diff, Stability) that manipulates structured AppState JSON instead of raw code — solving the reliability problem in AI code generation. Zero to live product in one week.
Led a team to build an app that estimates pet age from a single photo. Full pipeline from model training to deployment: EfficientNet-B0 (PyTorch → ONNX) served via FastAPI, with Supabase + Cloudflare R2 for storage and a Flutter frontend.
Flutter app with Gemini-based AI travel recommendations, Kakao OAuth2 authentication, Naver Map SDK route previews, and async API layers with retry and error handling. Built with GetX MVVM architecture.
React + TypeScript dashboard for visualizing vehicle driving data. Zustand state management, TanStack Query caching, MSW mock APIs for independent frontend development, and Recharts for data visualization.
Flutter app that tracks caffeine intake and predicts optimal sleep timing based on user biometrics. Interactive half-life decay visualization with weekly and monthly stats. 5-person cross-university team.
Dart, TypeScript, JavaScript, Python, HTML/CSS
Flutter, React, Next.js, FastAPI, Node.js
OpenAI API, Gemini, EfficientNet-B0, Multi-Agent Architecture
Docker, Jenkins, Cloudflare, Supabase, Vercel, PostgreSQL
6-month intensive in full-stack development, applied AI, and startup building. Working daily in English with native-speaking peers and mentors.