Xiaoxiang Cheng

Research Associate, Data Science, Machine Learning Applications at Ansteel Beijing Research Institute Co., Ltd.

Processed steel tabular data through imputation, cleaning, and normalization using Python (Pandas, NumPy) to improve data quality for downstream analysis and modeling. Built automated data pipelines for steel image preprocessing (cropping, resizing, annotation) using OpenCV to create high-quality datasets for defect analysis. Applied feature selection (Random Forest) to rank feature importance and correlations for dimensionality reduction and identifying top features for product metrics. Designed machine learning models using tabular data for quality metrics prediction and defect classification, enabling automated prediction. Developed deep learning models for steel surface defect detection using image data (CNN-based classifiers and object detection approaches). Queried and analyzed large-scale steel manufacturing datasets in MySQL for batch ratios, hourly process measurements, and product metrics, enabling machine learning and predictive modeling. Built and maintained interactive Pow

Mesh Data Processing Intern at Shining 3D Technology Co., Ltd.

Developed and optimized a remeshing method in C++ for triangular mesh processing, generating high-quality 3D dental mesh data; supported reliable visualization and analysis.