π― AI Backend Engineer Β· Search / RAG Systems
π Sejong University, Intelligent Mechatronics Engineering
π°π· Based in Seoul Β· π―π΅ Learning Japanese (JLPT N3)
I design and operate production AI search systems end-to-end β from incremental ETL pipelines (Airflow + watermark) to hybrid retrieval (Elasticsearch BM25 + Qdrant dense vector), Neo4j-based GraphRAG, and async LLM pipelines (asyncio.Semaphore + ainvoke).
Built and operated these systems during 13 months of military service in the ROK Air Force Data Platform Team.
Republic of Korea Air Force Β· Data Engineering Soldier
Intelligence Information System Management Group, Data Platform Team
Apr 2025 β May 2026
- π News Hybrid RAG + GraphRAG Search System β Airflow ETL Β· 3-signal Hybrid Retrieval (ES + Qdrant) Β· Neo4j GraphRAG Β· async LLM extraction
- ποΈ Internal Data Hub β Kotlin Spring Boot Β· KotlinJDSL Β· event-driven RDBβES sync Β· React + TS full-stack
- π¬ Real-time Meeting Text Analytics β Spring/FastAPI split Β· Kiwi NLP Β· dual-token JWT (key_version + SHA-256/AES-GCM)
π§ yh47529722@gmail.com
π LinkedIn
π§ Currently building: open-source news hybrid search benchmark with retrieval evaluation (Recall@K, MRR, nDCG)
