I am currently pursuing a Master of Science in Computer Science at Columbia University, with a focus on Machine Learning, Artificial Intelligence, and Natural Language Processing. I received both Bachelor of Science in Intelligent Mechatronics Engineering and Bachelor of Engineering in Data Science from Sejong University in February 2024.
Machine Learning, Artificial Intelligence, Data Science, Natural Language Processing (NLP), Multimodal Processing
I worked as a Research Assistant at the Mobile Intelligent Embedded System Laboratory at Sejong University from September 2021 to March 2024, under the supervision of Professor Hyung Seok Kim. During this time, I led several projects in AI and embedded systems, focusing on multimodal emotion recognition, edge computing, and human-robot interaction.
| No. | Title | Status |
|---|---|---|
| 1 | Eesun Moon, A.S.M Sharifuzzaman Sugar, Hyung Seok Kim. "Multimodal Daily-life Emotional Recognition Using Heart Rate and Speech Data from Wearables." IEEE Access, vol. 12, pp. 96635-96648, 2024. DOI | Published in IEEE Access |
| 2 | Taein Kim, Eesun Moon, Hoyeon Kang, Hyung Seok Kim. "OMER-NPU: On-device Multimodal Emotion Recognition on Neural Processing Unit for Low Latency and Power Consumption." Neural Computing and Applications. In submission | Peer Review |
| 3 | Eesun Moon, Hyungseok Kim. "Multi-modal Emotion Recognition Using Physiological Sensor and Speech." In Proceedings of the 38th Annual Conference of ICROS 2023. DOI | Published in ICROS2023 |
| 4 | A. S. M. Sharifuzzaman Sagar, Samsil Arefin, Eesun Moon, Md Masud Pervez Prince, L. Minh Dang, Hyung Seok Kim. "A Gaussian Process-Enhanced Non-Linear Function and Bayesian Convolution–Bayesian Long Term Short Memory Based Ultra-Wideband Range Error Mitigation Method for Line of Sight and Non-Line of Sight Scenarios." Mathematics, vol. 12, no. 23, pp. 3866, 2024. DOI | Published in Mathematics |
- Designed and implemented a personalized support message generation pipeline leveraging LangChain, OpenAI, and RAG algorithms to optimize message relevance and accuracy through prompt-tuning techniques
- Streamlined deployment with Docker on AWS EC2, integrating with Next.js front-end server for live photo booth events
- Delivered system to 500+ attendees at professional soccer stadiums, achieving 20% improvement in user satisfaction
Research Assistant 📎
- Led multimodal emotion recognition projects for government and corporate initiatives, focusing on solutions for on-device AI, using Python, TensorFlow, and MongoDB on Linux
- Optimized deep learning models by integrating multiple data modalities (heart rate, electroencephalogram, speech, and image) through score-based fusion, achieving 99.68% classification accuracy without increasing network complexity
- Embedded and deployed ONNX-optimized models onto Mobilint’s MLA100 NPU, reducing average power consumption by 3.12 times and latency by 1.48 times for edge computing
- Published papers in IEEE (Institute of Electrical and Electronics Engineers) and NCAA (Neural Computing and Applications), presented a poster at ICROS (Institute of Control, Robotics, and Systems), and conducted real-time deployment demonstrations at KIST (Korea Institute of Science and Technology)
Database ETL of NYC Crime Analysis (Sep 2024 – Dec 2024) 📎
- Led development of a comprehensive database system to analyze NYC crime data by integrating two public datasets with 14 million+ raw records, utilizing PostgreSQL for data management and relational mapping
- Preprocessed raw datasets to resolve inconsistencies, designed custom ER diagram, and implemented relational tables in PostgreSQL, enabling execution of 10+ complex queries for in-depth crime pattern analysis
Outfit Coordination Recommender System (Aug 2023 – Sep 2023) 📎
- Developed generative AI-driven outfit recommender by fine-tuning KoAlpaca LLM with PEFT-LoRA for personalized suggestions
- Constructed fine-tuning dataset of 28,000 Q&A pairs using KoNLPy, achieving an 80% satisfaction rate from faculty and peers compared to previous models
Spam Detection on Social Networking Services (Mar 2022 – Jun 2022) 📎
- Spearheaded NLP-based spam detection project to classify social media posts as potential advertisements, addressing scalability in unstructured data, and awarded 1st place in graduation project competition
- Automated data collection from social media platforms using Selenium, analyzed textual data, and implemented ranking algorithms to sort posts by decreasing the likelihood of being advertisements, achieving 0.8 cosine similarity for classification accuracy
Stock Sentiment Analysis and Quantitative Modeling (Sep 2021 – Mar 2022) 📎
- Correlated stock prices with financial statements from 2020 to 2021 for 50 companies, building quantitative models using PyTorch and comparing linear regression with LSTM for stock trend predictions
- Built a sentiment dictionary by extracting keywords from economic news articles using BeautifulSoup, analyzing stock price fluctuations within three days of publication to assess impact of keyword sentiment scores on stock prices