Jae Young Kim Email and Phone Number

• Developed template-based RSA headline/description generation model using multiple NLP models.• Developed keyword recommendation model suggesting relevant keywords to help 3rd party sellers to update item title.• Fine-tuned Clip model with product image, category pairs to develop image retrieval system.

• Researching incorporating qualifiers to Temporal Knowledge Graph prediction.• Working in VENICE project converting Telegram and News message to Knowledge Graph.

· Worked with AI engineers in Personalization Technology Team to develop recommender system of IPTV kids platform.· Developed recommender system using Neural Collaborative Filtering and GRU4REC model.· EDA of watch log data and grouped users based on watching pattern of the users.· Reviewed recent deep learning based recommender system papers such as SRGNN and HRNN and reported summary.

· Create recommender system with user visit logs using Matrix Factorization and Deep Learning.· Visualized domestic real estate transaction volume into a graph and detected communities.· Created Indicators to evaluate stock portfolio stability.

· Designed architecture of database using MySQL. Designed visualization of the performance of content partners on the webpage· Worked with software engineers to launch a new version of online settlement system storing the revenue from platform companies, issuing invoice automatically to the content partners based on the contract and visualizing profit, total content viewing time.

• Assisted research on development of motion detection algorithms to create rehabilitation-facilitating robots and other medical equipment. • Applied data clustering methods such as SVM (Support Vector Machine) to prototypes of robots that would be used to support limb mobility. I then analyzed sensory information (from person wearing the robot) detected by SVM to find that the precision of the data achieved 93% accuracy. Conducted multiple experimentation rounds with SVM prototypes on… Show more • Assisted research on development of motion detection algorithms to create rehabilitation-facilitating robots and other medical equipment. • Applied data clustering methods such as SVM (Support Vector Machine) to prototypes of robots that would be used to support limb mobility. I then analyzed sensory information (from person wearing the robot) detected by SVM to find that the precision of the data achieved 93% accuracy. Conducted multiple experimentation rounds with SVM prototypes on stroke victims to further assess the effectiveness of products.• Co-researched method of developing an AI personal trainer that would be able to automatically control treadmill velocity for physical therapy patients. Applied machine learning algorithms such as SVM and DNN (Deep Neural Network) and compared accuracy between different modes. Show less

National Assembly Police Security Guard.Served military service as the National Assembly Guard at the National Assembly.