91% faster response time | 90% faster incident detection | Open Source Contributor
Backend Developer (3Y 7M) → AI/MLOps Engineer
문제를 발견하면 그냥 지나치지 않는 개발자입니다.
- 모니터링 시스템이 없으면? → 직접 만들어서 오픈소스 기여 (Scouter Contributor)
- 성능이 느리면? → 3단계 최적화로 응답시간 91% 개선 (1,400ms → 120ms)
- 반복 작업이 불편하면? → Python 2주 독학 후 사내 150명이 쓰는 자동화 툴 개발
현재는 백엔드 최적화 경험을 AI/ML 서비스로 확장하고 있습니다.
- 📈 80,000 concurrent users game service response time improved by 91% (1,400ms → 120ms)
- 💰 Monthly cloud cost reduced by 30% through Redis optimization & infrastructure scale-down
- ⚡ Incident detection time reduced by 90% (30min → 3min) with custom Scouter APM plugin
- 🤝 Open Source Contributor: Merged MS Teams alert plugin to Scouter official repository
- 📊 Log analysis time reduced by 83% (30min → 5min) with Azure Log Analytics + KQL
- 🔧 Developed internal CDN automation tool (Python) → 83% work time reduction (3min → 30sec)
- 🚀 Built end-to-end ML pipeline with Airflow (crawling → preprocessing → training → serving)
Performance optimization through 3-stage caching strategy
- Redis caching + HTML static cache → 91% response time improvement
- Apache JMeter load testing for 80,000 concurrent users
- Monthly cloud cost reduction: 30%
Real-time monitoring system for 25 servers
- Developed custom Teams webhook alert plugin
- Merged to official Scouter repository (Open Source Contributor)
- Incident detection time: 30min → 3min (90% reduction)
Eliminating repetitive manual work
- Python tkinter GUI tool (learned Python in 2 weeks)
- 83% work time reduction (3min → 30sec)
- Used by 150+ employees company-wide
End-to-end ML automation
- FastAPI model serving + MLflow experiment tracking
- Airflow DAG: crawling → preprocessing → training → deployment
- Web UI for non-technical team members to run experiments
