About

Hi! I'm Jiyoung Park. I'm working at NAVER corp as a Research Engineer.

Keywords

  • Music Information Retrieval, Music Recommendation, Deep Learning, Audio Signal Processing
  • Scala, Spark, Python, Hive, Shell Script, SQL, Flask, Hadoop, Oozie, Go, Presto, RabbitMQ, Zeppelin

Education

  • [2016-2018] M.S. in Graduate School of Culture Technology
    Music and Audio Computing Lab, KAIST, Korea. (Advisor: Juhan Nam)
  • [2012-2015] B.S. in Electronic Physics and B.E. in Electronics Engineering
    HUFS, Korea. (Early Graduation, Double Major)

Experience

  • [2018.02~present] Research Engineer, NAVER corp., Korea.
  • [2017 Summer] Research Intern, NAVER corp., Korea.
  • [2015 Summer] Research Intern, Music and Audio Research Group, Seoul National University, Korea.

Industrial Experience

  • Personalized Playlist Recommendation
  • - Personalized playlist according to mood/genre.
    - Applying a hybrid of collaborative/content-based filtering.
    - Provides a fresh list of songs while reflecting personalized tastes in daily/weekly updates.
    - Spark, Scala, Hive, Oozie
  • Data Pipeline
  • - Automated data pipeline to extract various features (tempo, beat, mood, high-level audio feature from a pre-trained model etc.) from audio including the latest released one.
    - Clustering same tracks among millions of songs using meta data and high-level audio features.
    - Python, RabbitMQ, Hive, Consul & Nomad, Shell Script, Pyarrow
  • Keyword-based Recommendation
  • - Recommendation songs by situation or AI speaker request from user.
    - Applying a hybrid of zero-shot model and matrix factorization.
    - Natural language processing to extract the main keyword from query.
    - Keras (backend Tensorflow), Python, Spark
  • Related Item Search
  • - Similarity search between millions of items.
    - Audio-based item-to-item recommendation (applying my ISMIR 2018 research).
    - Annoy, Faiss, Spark

Publications, Demos and Technical Reports

Google Scholar
  • Representation Learning Using Artist Labels for Audio Classification Tasks
    Jiyoung Park, Jongpil Lee, Jangyeon Park, Jung-Woo Ha and Juhan Nam
    Music Information Retrieval Evaluation eXchange (MIREX) in the 18th International Society for Musical Information Retrieval Conference (ISMIR), 2017
    (1st place in the Music Mood Classification task across all algorithms submitted so far)

Contact

gyung527@gmail.com
j.y.park@navercorp.com
LinkedIn Google Scholar