|
“My best ideas come from my daily walks.” I build practical AI products from problems I personally encounter: learning, translation, technical writing, and speaking practice. I enjoy turning an idea into a small but operational system by connecting the backend, data pipeline, AI workflow, user interface, evaluation, deployment, and technical documentation. I turn “I wish this existed” into running software. |
|
Languages
Product & Applications
AI & Data
Infrastructure & Operations
| Project | What it demonstrates | Stack | Status |
|---|---|---|---|
| Algorithm RAG Engine | A personalized algorithm learning system evolving from cross-lingual retrieval toward weakness analysis, recommendation, and structured review loops. | Python OpenAI Embeddings NumPy Slack Docker S3 GitHub Actions |
Active iteration |
| hunbot.dev | An AI-native technical blog with multilingual publishing, a local LLM translation workflow, and operational monitoring. | Astro MDX Ollama Python Vercel Grafana |
Live |
| local-llm-translator | A Korean MDX translation pipeline supporting local and cloud inference, quality scoring, PostgreSQL experiment tracking, and an operations dashboard. | Python FastAPI Ollama RunPod PostgreSQL pgvector SvelteKit |
In development |
| Project | Description | Stack | Link |
|---|---|---|---|
| BKGA Extension | A VS Code Korean grammar assistant for Markdown technical writing, with diagnostics, hover explanations, quick fixes, and a custom dictionary workflow. | TypeScript VS Code API Bareun API |
Repo ↗ / Blog ↗ |
| Topic | Note |
|---|---|
| Algorithm RAG Update | Rethinking the project from a problem recommender into a personalized algorithm learning system. Read ↗ |
| Local LLM Translation | Experiments in translating Korean technical posts while preserving MDX structure and evaluating translation quality. Read ↗ |
| Japanese Speaking AI | Designing a real-time Japanese conversation trainer with voice interaction, RAG, and structured feedback. Read ↗ |