Role Overview
Developed multi-class machine learning classifiers to streamline root cause analysis of wafer defects, accelerating the debugging process for semiconductor manufacturing.
Technical Work
Machine Learning Classifiers
Built and evaluated multiple classification approaches:
- Support Vector Machines (SVM)
- K-Nearest Neighbors (KNN)
- XGBoost gradient boosting
- Convolutional Neural Networks (CNNs)
Feature Engineering
- Engineered training features using Pandas, NumPy, and scikit-learn
- Applied online learning techniques where applicable
- Evaluated performance with cross-fold validation
Recognition
Selected as a top candidate to travel to Samsung headquarters in Korea for workshops on cross-functional collaboration, representing the intern cohort at the executive level.
Impact
Automated defect classification reduced manual inspection time and improved yield analysis, contributing to faster production cycles and cost savings.