Backend and full-stack developer focused on real services, measurable performance improvements, and practical AI tools.
수치와 지표로 시스템의 가치를 증명하는 개발자입니다.
- Portfolio: https://rocky-limburger-416.notion.site/Jang-Hyeong-27407c0007d3821a8f420155ea7fcd7e?pvs=143
- Blog: https://coldmans.tistory.com/
- Solved.ac: https://solved.ac/profile/jinhinjala26
- Email: jangboss02@gmail.com
| Area | What I build |
|---|---|
| Backend | Spring Boot APIs, JPA data modeling, PostgreSQL/MySQL, REST contracts, performance tuning |
| Full-stack service | React/Vite/Next.js frontends connected to real backend APIs |
| AI tools | LLM API integrations, MCP servers, agent workflows, multimodal demos |
| Operations | Docker, GitHub Actions, k6 load testing, README-first documentation |
| Project | What it shows |
|---|---|
| INU Timetable | Spring Boot timetable service with Gemini-assisted PDF parsing, PostgreSQL, k6 performance tuning, and real student usage |
| INU Timetable Front | React/Vite timetable UI with course search, wishlist, generated combinations, timetable visualization, and export flows |
| codex-diary | Local macOS app and CLI that turns Codex Chronicle summaries into daily work reports and diary-style reflections |
| LastGuardianMCP | MCP server that estimates late-night transit last-departure times using Google Routes API and binary search |
| K-Context Guide | Korean-context multimodal demo with Spring Boot, Next.js, WAV recording, and an OpenAI-compatible Kanana API flow |
| GAIA Capstone | Agentic browser QA platform that turns product specs into executable test flows and evidence-based reports |
- How I improved INU Timetable API performance from multi-second failures to stable sub-second responses under k6 load.
- How JPA entity design, fetch strategy, pagination, indexing, and connection pool settings affected real API behavior.
- How to connect React UI state, backend REST APIs, authentication, wishlist state, and timetable conflict logic.
- How MCP tools expose real-world APIs to LLM clients, including cost and reliability tradeoffs.
- How AI demos should handle fallback paths, privacy boundaries, and inspectable outputs instead of only model calls.
Backend: Java, Spring Boot, JPA, PostgreSQL, MySQL, Redis, FastAPI, Python
Frontend: React, Vite, Next.js, TypeScript, JavaScript, Tailwind CSS
AI/Automation: OpenAI-compatible APIs, Gemini API, MCP, Playwright, Codex CLI
Infra/Testing: Docker, GitHub Actions, k6, JUnit 5, MockMvc, Swagger/OpenAPI