Chan Kim Email & Phone Number

I'm interested in NLP, especially LLM.Areas of Research Interest- LLM Evaluation (Aligning with human judgment)- Input Refinement (ex. Query Rewriting, Reranking of Retrieved Passages)Career Path- [Upstage] AI Solution Reliability Engineer Intern (2024.05 ~ Present)- [Chung Ang University] LILAB NLP Research Intern (2024.01 ~ 2024.04)- [DIYA] 5th, AI society (2023.01 ~ Present)- [Dacon] Data Scientist (2023.05 ~ 2023.12)- [Google] Machine Learning BootCamp (2023.09 ~ 2023.11)- [OSSCA] PyTorch and PyTorch Lightning (2023.07 ~ 2023.09)- [BoostCamp AI Tech] 4th NLP Track competition (2022.09 ~ 2023.02)You can refer to my GitHub and blog.- https://github.com/chanmuzi- https://chanmuzi.tistory.com/Below is a list of projects I engaged in through studying and at work.1. RAG-basedLaw Specialized Chatbot Model / Chung Ang University - Mar. 2024 ~ April. 2024- Establishing baselines for instruction tuning of LLMs- Researching evaluation methodologies for LLMs2. Development of Educational Chatbot Services Using LLM / Dacon - Oct. 2023 ~ Dec. 2023- Led the integration of chatbots in machine learning educational content- Creating AI services for quantitative assessments and qualitative feedback, incorporating API-based solutions3. AI vs. Human Text Discrimination Competition Design & Management / Dacon - Sep. 2023 ~ Nov. 2023- Initiated and managed a competition on AI-generated content detection using the OpenAI API- Crafted a novel dataset by modifying colloquial Twitter texts for diverse expressions- Established a baseline and explored various solution strategies4. Visual Question Answering (VQA) / Dacon - Jun. 2023 ~ Aug. 2023- Orchestrated a VQA competition, handling dataset preprocessing and competition structuring- multi-modal task, accuracy metric5. Document Visual Question Answering (DocVQA) - Dec. 2022 ~ Feb. 2023- Collaborated in a corporate-linked project with UpStage- Addressed the challenge of aligning visual (bbox) and linguistic information (text) in DocVQA tasks- Developed attention maps to visually represent model focus areas, addressing weaknesses in diagram and figure interpretationIn addition to the above list, I have also undertaken the following NLP tasks: - ChatGPT Prompt Engineering Competition- Open Domain Question Answering (ODQA)- building a Korean Law Dataset for Relation Extraction- Relation Extraction (RE)- Semantic Text Similarity (STS)These projects have significantly enhanced my foundational skills and knowledge in Natural Language Processing (NLP).

• 

  서울시립대학교 (University Of Seoul)

  영어영문학