Humoyunbek Abdukarimov
Data Scientist @ HumbleBeeAI · Applied ML, Behavioral Biometrics & Fraud Detection · Incheon, South Korea
Incheon, South Korea
Hi, I’m Humoyunbek — a Data Scientist at HumbleBeeAI in Incheon, South Korea, where I build applied machine-learning systems around behavioral biometrics, fraud detection, and time-series & blockchain analytics.
Over the past couple of years at HumbleBeeAI I’ve grown from Data Analyst to Head of Data Science to Data Scientist — founding the data-science team, leading a group of seven, and shipping work like behavioral bot-detection pipelines (keystroke and mouse dynamics), on-chain fraud detection across Cronos and Ethereum, investor-churn prediction for fintech, and time-series forecasting for offshore engineering. I work mostly in Python, SQL, scikit-learn, LightGBM, and BigQuery.
I’m also an honours student in Integrated System Engineering at Inha University and a recipient of the Global Scholarship 2.
Outside of work, I like writing — you’ll find tech notes and life thoughts on the blog, and a bookshelf where I keep track of what I’m reading and what I made of it. Feel free to reach out.
news
| Aug 28, 2026 | Our paper “Training-Free Few-Shot Personalization for Mobile Mood Inference with TabPFN” was accepted to the DiversityOne Challenge Workshop at ACM UbiComp/ISWC 2026. |
|---|
latest posts
| Jul 18, 2026 | When the World Curves |
|---|---|
| Jun 30, 2026 | Benchmarking TabPFN V2 against XGboost and Catboost |
selected publications
- Training-Free Few-Shot Personalization for Mobile Mood Inference with TabPFNIn DiversityOne Challenge Workshop, ACM International Joint Conference on Pervasive and Ubiquitous Computing and International Symposium on Wearable Computers (UbiComp/ISWC), 2026Accepted; to appear in the ACM Digital Library