Building AI products that matter.
CTO @ Rivetta — Designing context engineering architectures for AI-powered K-webnovel translation. While many claim LLMs have "solved" translation, real-world demands prove otherwise: readers need cultural nuance and domain expertise that raw AI cannot deliver. I architect systems that bridge this gap.
CEO @ AlphaView — AI-driven stock news analysis platform. (Solopreneur)
26.02 Official Launch / Reached MAU 20k
|
CTO — K-Webnovel AI Translation Translating Korean webnovels for the US market. LLMs alone fail to satisfy readers—lacking domain knowledge and cultural understanding. I design context engineering architectures that deliver translations readers actually want. |
CEO — AI Stock Analysis Building alphaview.kr, an AI-powered platform that analyzes stock news and market signals to help investors make informed decisions. |
| Experience | Focus | |
|---|---|---|
| 🧬 | Neuroscience | Trans-synaptic retrograde signaling research |
| 📚 | 8 Years Literary Translation | Novels, webtoons, creative content |
| 🔧 | 3 Years Technical Translation | Documentation, ML papers, technical specs |
"This domain expertise informs my approach to AI systems that truly understand context—something raw LLMs consistently fail to achieve."
GPU Hill Climbing Ensemble for NVIDIA Merlin Framework
CTR prediction on 10M rows of real financial data (Recommender System)
📄 Technical Report (EN) | (KR)
-Top_0.01%25_(%2324)-00d084?style=flat-square){kind=icon}
Recommender System CTR Prediction NLP Prompt Engineering Time Series Computer Vision Tabular ML Anomaly Detection
🔥 GitHub Trending #1 · 🛡 Collaborator (write access) · sorted by merged-PR count
AI coding-agent runtimes · orchestration · hooks · skills · memory
| Project | Focus | ⭐ Stars | Merged |
|---|---|---|---|
| 🔥 🛡 oh-my-claudecode | Teams-first multi-agent orchestration for Claude Code | 29.5k |
 |
| ouroboros | Specification-first workflow engine — Stop prompting. Start specifying. | 2.4k |
 |
| 🔥 oh-my-codex | Hooks, agent teams, and HUDs for Codex CLI | 23.7k |
 |
| multica | Managed agents platform — coding agents as teammates | 14.9k |
 |
| 🔥 nanobot | Ultra-lightweight personal AI agent (~4K lines) | 39.8k |
 |
| agent-skills | Production-grade engineering skills for AI coding agents | 16.7k |
|
| agentmemory | #1 persistent memory for AI coding agents | 1.7k |
|
| alphaclaw | Setup harness for OpenClaw — deploy in minutes, no CLI | 1.2k |
|
LLM app platforms · recommender systems · voice synthesis
| Project | Focus | ⭐ Stars | Merged |
|---|---|---|---|
| dify | Production-ready platform for agentic workflow development | 138k |
 |
| voicebox | Open-source voice synthesis studio | 19.3k |
|
| gorse | AI recommender system (classical + LLM rankers) | 9.6k |
|
RF / DSP / hardware
| Project | Focus | ⭐ Stars | Merged |
|---|---|---|---|
| PLFM_RADAR | Open-source low-cost 10.5 GHz PLFM phased array RADAR | 12.2k |
 |
| Platform | Models | Share | Messages | Tokens | Cost |
|---|---|---|---|---|---|
 | opus-4-6 sonnet-4-6 sonnet-4-5 +3 | ████████████████░░░░ 78.4% | 327,629 | 29.0B | $24,841 |
 | 5.4 5.3-codex 5.3-codex-spark +1 | ████░░░░░░░░░░░░░░░░ 21.6% | 120,878 | 17.3B | $6,833 |
 | 2.5-pro 2.5-flash | █░░░░░░░░░░░░░░░░░░░ 0.0% | 19 | 540.4K | $0.40 |
| Total | 448,526 | 46.3B | $31,674 | ||
Token Composition
-43.3B-2ECC71?style=flat-square){kind=icon}
-2.0B-27AE60?style=flat-square){kind=icon}
-931.5M-3498DB?style=flat-square){kind=icon}
-176.5M-9B59B6?style=flat-square){kind=icon}
Languages & ML
Backend & Data
Frontend
DevOps & Testing