NLP AI Research Engineer

I am Seunghyun Ji, currently studying NLP AI since my undergraduate years, without pausing after earning a master's degree. Despite the rapid changes in AI technology, my focus has been on studying domain-agnostic AI model methodologies that go beyond short-term solutions, thus enabling the implementation of customized solutions quickly according to requirements. My strengths include:

• Ability to follow the latest research trends and perform meta-analysis

  • NeurIPS 2023 Oral/Spotlight paper 리뷰
  • SIGUL 2024 workshop (in LREC-COLING 2024) **Program committee,** Reviewer of EMNLP 2023, ACL 2023

• Intuitive understanding of concepts for exchanging opinions with people from various fields

  • 인프런 무료 강의 활동

• Participation in multiple AI model implementation projects for various NLP tasks (Conditional Generation, Classification, Text Generation), including experiences in implementing and training Vision, Speech, and Multimodal models using general ML knowledge

Contact Information

• Email: [email protected]
• Github: https://github.com/comchobo
• Blog: https://blog.naver.com/sorryhyun96

Work Experience

• 2024.03 ~ A-ha Co., Ltd. (be expected)

  NLP AI Research Engineer (Full-time)

• 2023.11 ~ 2023.12 IBS Co., Ltd. (based in Japan)

  NLP AI Research Engineer (Part time job)

  • Writing AI part of the project plan for improving existing systems
  • Review of the latest LLM research trends
    • Portfolio: Review of accepted paper for NeurIPS 2023 (in Korean)
      • I made an underestimation due to misinterpretation for the “Unpaired Multi-Domain Causal Representation Learning” on page 28.
  • Proposal for collaboration pipeline between humans and AI and efficiency improvement of work pipeline

• 2023.03 - 2023.10 TelePIX Co., Ltd. (based in Korea)

  NLP AI Research Engineer (AI Research Team, DS Department, Full time job)

  • Implementation of an unstructured text processing pipeline that satisfies ontology (to improve the performance of prediction models)
  • Implementation and performance improvement of various models such as filtering, translation models, and sentence embedding
    • Structured pipeline code: text_to_triplet
    • Translation model code: Lora training for domain-adapted NLLB Machine translation model
    • Achievement of submitting a methodology paper* on data collection on SIGUL 2024 workshop in LREC-COLING 2024, and patent application.
  • Design and implementation of evaluation pipeline for audio-text multimodal models
    • Portfolio of audio-text multimodal model evaluation pipeline: Code diagram summary
    • Implementation within 2 weeks and performance improvement compared to existing unimodal models
  • Other responsibilities include implementing vision models for cloud detection, participating in the latest research seminars, etc.
    • Portfolio of research proposal seminar: Event search system using satellite photos (in Korean)
    • Portfolio of technical problem interview substitute proposal: Example of AI technical problem

  *. The paper will be disclosed via Arxiv soon.