Koh Geonho
Junior ML Engineer
Seoul
Summary
Machine Learning Engineer with strong experience in developing recommendation models and scalable backend systems. Skilled in end-to-end product development, from model design to deployment, with a proven ability to collaborate cross-functionally and drive measurable business impact. Known for bridging the gap between ML and engineering, delivering solutions that are both technically robust and user-focused.
Overview
3
3
years of professional experience
7
7
years of post-secondary education
Work History
ML Engineer
BucketPlace
02.2023 - Current
- Substitutional Goods Recommendation Model Development
Improve collaborative filtering model by integrating GNN with product metadata and graph random-walk model for candidate generation, leading to personalized and popular item recommendations.
Impact: +12.5% GMV from product pages, +12% conversion rate. - Keyword Module & Feed Development
Built pipelines for personalized keyword recommendation using session-based models and LLM embeddings; developed keyword feed displaying related content.
Impact: +21% tag clicks/user, +32% content views. - New Content Recommendation System
Implemented content-based recommendation using KELIP embeddings and similarity search for personalized discovery of new content.
Impact: +54% new content exposure, +2.5% views/user. - Viewer Feed Launch
Expanded recommendation APIs by developing taxonomy classification model using multimodal embeddings to enhance user exploration/exploitation experience.
Impact: +13.6% post-search content views, +2.7% view-to-click conversion. - Global Personalized Feed Launch
Developed end-to-end pipelines for global regions, including Explore, Following, and Interest feeds; managed full cycle from model to deployment.
Impact: Successfully localized recommendation systems across regions. - Discovery Service Launch
Backend development for metadata pipeline design by participating in API-based recommendation service development for real-time content queries.
Data ML Engineer Intern
BucketPlace
12.2021 - 08.2022
- User Embedding for Demographic Prediction
Generated user embeddings from user behavioral logs using BERT; outperformed traditional matrix factorization but not deployed due to training cost. - Feature Enhancement for Product Ranking Model
Improved model performance by adding features such as device type, discount rate, and review count, etc. - How-to Recommendation Model Experimentation
Tested multiple recommendation models (Item-CF, session-based, etc.) to optimize related content suggestions.
Impact: +90% recommendation coverage, +5% CTR. - Review Sentiment Analysis
Conducted aspect-based sentiment analysis on product reviews using KoBERT as part of a university-affiliated project.
Education
Bachelor of Science - School of Computing
KAIST
Daejeon, South Korea 05.2016 - 03.2023
Skills
- Machine Learning
- NLP
- Recommendations
- Go
- PySpark
- Python
- PyTorch
- Tensorflow
- SQL
- NoSQL
- AWS
Timeline
ML Engineer
BucketPlace
02.2023 - Current
Data ML Engineer Intern
BucketPlace
12.2021 - 08.2022
Bachelor of Science - School of Computing
KAIST
05.2016 - 03.2023
Similar Profiles
Jiwon SongJiwon Song
Software Engineer, Backend at BucketplaceSoftware Engineer, Backend at Bucketplace
Yewon JeongYewon Jeong
Beauty Advisor Assistant at ChanelBeauty Advisor Assistant at Chanel
Jung Bin KimJung Bin Kim
Associate Consultant Trainee at Bain & CompanyAssociate Consultant Trainee at Bain & Company