Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I'm Kunhong Yu, a data scientist and multimodal ML researcher with a track record of turning messy data into actionable insights. I design scalable data pipelines, build end-to-end ML systems, and explore vision-language and NLP problems across fintech, e-commerce, and research settings, collaborating with cross-functional teams worldwide. I enjoy translating complex data into practical solutions, deploying models to production, and mentoring others on best practices for experimentation, reproducibility, and responsible AI. I’m excited to apply my skills to real-world challenges in data engineering, ML research, and product development.…I'm Kunhong Yu, a data scientist and multimodal ML researcher with a track record of turning messy data into actionable insights. I design scalable data pipelines, build end-to-end ML systems, and explore vision-language and NLP problems across fintech, e-commerce, and research settings, collaborating with cross-functional teams worldwide. I enjoy translating complex data into practical solutions, deploying models to production, and mentoring others on best practices for experimentation, reproducibility, and responsible AI. I’m excited to apply my skills to real-world challenges in data engineering, ML research, and product development.

Kunhong Yu

Data Scientist, AI Engineer, Data Scraping Specialist, +1





I'm Kunhong Yu, a data scientist and multimodal ML researcher with a track record of turning messy data into actionable insights. I design scalable data pipelines, build end-to-end ML systems, and explore vision-language and NLP problems across fintech, e-commerce, and research settings, collaborating with cross-functional teams worldwide. I enjoy translating complex data into practical solutions, deploying models to production, and mentoring others on best practices for experimentation, reproducibility, and responsible AI. I’m excited to apply my skills to real-world challenges in data engineering, ML research, and product development.…I'm Kunhong Yu, a data scientist and multimodal ML researcher with a track record of turning messy data into actionable insights. I design scalable data pipelines, build end-to-end ML systems, and explore vision-language and NLP problems across fintech, e-commerce, and research settings, collaborating with cross-functional teams worldwide. I enjoy translating complex data into practical solutions, deploying models to production, and mentoring others on best practices for experimentation, reproducibility, and responsible AI. I’m excited to apply my skills to real-world challenges in data engineering, ML research, and product development.

Available to hire

I’m Kunhong Yu, a data scientist and multimodal ML researcher with a track record of turning messy data into actionable insights. I design scalable data pipelines, build end-to-end ML systems, and explore vision-language and NLP problems across fintech, e-commerce, and research settings, collaborating with cross-functional teams worldwide.

I enjoy translating complex data into practical solutions, deploying models to production, and mentoring others on best practices for experimentation, reproducibility, and responsible AI. I’m excited to apply my skills to real-world challenges in data engineering, ML research, and product development.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Fluent

Polish

Beginner

Work Experience

Web Scraping Data Engineer at Point72 Poland

September 1, 2025 - September 1, 2025

Developed and optimized web scraping pipelines using Python, BeautifulSoup, and Playwright to extract data from JavaScript-heavy pages and APIs, ensuring compliance with legal guidelines. Led migration of CI/CD pipelines from Jenkins to Airflow, designed DAGs, and built custom operators to integrate scraping tasks with analytics systems. Incorporated large-language-model-based processing to extract and structure data from images, PDFs, and long texts.

Deep Learning Researcher at VSPN

February 1, 2021 - February 1, 2021

Detected main actions in phone screen images with YOLOv3 and performed text recognition with CRNN OCR. Converted trained model to ONNX for production deployment.

Machine Learning Researcher/Data Scientist at Kingstar Fintech

December 1, 2020 - December 1, 2020

Developed end-to-end CRNN OCR models to detect bounding boxes and recognize text on receipts. Improved text clarity using AutoEncoder and distillation for smaller models and distributed training. Applied Conditional GANs to generate augmented datasets for image captcha recognition; built multi-task models using GoogLeNet and MobileNetv2.

Machine Learning Engineer Intern at PhoenixNet

September 1, 2018 - September 1, 2018

Cleaned and preprocessed user clickstream data; tuned hyperparameters of deep learning recommendation models; improved training efficiency.

Web Scraping Data Engineer (Project: Japanese Job Posting) at Point72

May 1, 2024 - May 1, 2024

Japanese Job Posting project: built scrapers and reusable frameworks to collect and process large-scale data from multiple Japanese job posting sites, with CI/CD pipelines using Jenkins and Airflow. Leveraged PySpark for large-scale analysis with data quality checks and time-series visualization.

Core Developer at Point72

May 1, 2025 - May 1, 2025

LLM Image and PDF Data Parsing: parsed PDFs from diverse sources using an internal GPT client to extract data from images, pure texts, and image-like tables. Implemented iterative improvement with chain-of-thought, in-context learning, and reflection mechanisms to refine accuracy.

Core Developer at UW

April 1, 2023 - April 1, 2023

Prompt Tuning by Context Template Pool Optimisation for Vision-Language Model: reviewed state-of-the-art literature, generated ideas with advisors, and developed a novel two-step algorithm for CLIP model prompt-tuning. Validated with large-scale experiments and disseminated results.

Core Developer at UW

June 1, 2022 - June 1, 2022

FullyLight: Keras-like neural network classifier implemented in R with a vectorized forward-backward pass; interactive visualization of training dynamics via Shiny.

Core Developer at Kingstar Fintech

May 1, 2020 - May 1, 2020

Anomaly Detection in Financial Transactions: preprocessed data with Python/SQL (Pandas/Sklearn/Matplotlib/Seaborn); evaluated multiple ML algorithms (logistic regression, Random Forest, AutoEncoder-based, XGBoost, CatBoost) to detect fraudulent activity; documented architecture and data handling in Markdown/Wiki.