Summary
Youyoung Jang is an AI/ML engineer based in Seoul with eight years of experience building production recommender systems and personalization pipelines for large consumer platforms. Currently at NAVER Shopping, she fine-tunes LLMs using user behavior histories and quantized product vectors, and developed an RQ-VAE based vector quantizer to extract semantic product IDs for generative retrieval. Previously at Toss she led end-to-end recommendation and two-stage ranking systems, designed multi-armed bandits for sparse, noisy feedback, and built graph embeddings and feature pipelines that materially improved engagement and conversion metrics. Her background in statistics and hands-on work across ads, commerce, and fraud detection gives her a strong blend of experimental rigor and production-first engineering. She frequently bridges modeling and data engineering—authoring Airflow operators, FAISS-based similarity pipelines, and scalable feature stores—so research ideas make it into live systems. An under-the-radar strength is her experience instruction-tuning LLMs for product keyword extraction, showing a knack for creative applied NLP in commerce contexts.
8 years of coding experience
5 years of employment as a software developer
Bachelor's degree, Statistics, Bachelor's degree, Statistics at 고려대학교