Tzu-Yuan Chen
M.S. Student · Seeking Data Science / ML Internship
chen.tzuyuan666@gmail.com
(+1) 945-371-1551 / (+886) 963-672-946
chentzuyuan.github.io
Dallas, TX / Taichung, Taiwan
M.S. student in Data Science (NCHU) and Social Data Analytics (UTD) with a B.S. in Applied Mathematics. Seeking a data science / ML internship to apply hands-on experience across 37 projects — including image classification (96.44% accuracy), distributed learning (150x speedup), and an ongoing thesis on flight delay prediction (AUC 0.93). Strong foundation in numerical analysis, deep learning, selective prediction, and end-to-end ML pipelines.
Technical Skills
Programming Languages
Python R SQL Qiskit Git LaTeXPython Ecosystem
pandas NumPy scikit-learn PyTorch Keras / TensorFlow statsmodels GeoPandas Selenium BeautifulSoup PyQt6ML / Deep Learning
ResNet EfficientNet U-Net Conditional GAN Autoencoder SVM KNN Random Forest MLP / DNN Kernel Ridge XGBoost Transfer Learning Focal LossMath & Theory
Linear Algebra / SVD HOSVD / Tucker Nyström Approximation Probability & Bayes Optimization Numerical MethodsBig Data & Distributed
Distributed SVM Communication-efficient SGD Kernel Approximation Selective Prediction Probability CalibrationLLM & AI
AI Agent Workflows Prompt Engineering RAG LLM API IntegrationGIS & Spatial
ArcGIS Python API GeoPandas ShapelyVisualization & Tools
ggplot2 Shiny Altair Matplotlib Qt Designer QuartoHardware & Platforms
IBM Quantum (IBMQ) Snowflake 3D Printing (FDM)Languages
Chinese (Native) English (Professional)Education
M.S. Social Data Analytics & Research (SDAR)
The University of Texas at Dallas — UTD-NCHU Dual-Degree Cohort
2025 – present
M.S. Data Science & Information Computing
National Chung Hsing University (NCHU), Taiwan
2024 – present
B.S. Applied Mathematics
National Chung Hsing University (NCHU), Taiwan
2020 – 2024
Projects
Computer Vision & Image Classification
AOI Defect Classification — ResNet-18 transfer learning on industrial inspection images (6-class). 96.44% accuracy
View Details →
View Details →
Cervical Cancer Screening — EfficientNet-B7 + Focal Loss on medical images (3-class). 86.1% avg accuracy
View Details →
View Details →
Handwritten Digit Recognition — 8-model comparison (Mean / SVD / HOSVD / SVM / KNN / RF / CNN) on USPS dataset. CNN 95.76%
View Details →
View Details →
Medical Image Analysis & Generation
Retinal Vessel Segmentation — U-Net (5-level encoder-decoder) + Focal Tversky Loss on DRIVE dataset. mIoU 0.351
View Details →
View Details →
Retinal Image Reconstruction — Convolutional Autoencoder for retinal fundus image reconstruction. PSNR 30.84 dB
View Details →
View Details →
Western Blot Image Synthesis — Conditional GAN (Generator + PatchGAN Discriminator) for biomedical image generation.
View Details →
View Details →
Scalable ML & Big Data
Nyström Kernel Ridge Regression — Gram matrix approximation on USPS digits (m=128). 20× speedup, 99.50% accuracy maintained
View Details →
View Details →
Distributed SVM — 5-worker parallelization with Smoothed Hinge Loss on a9a dataset. 150× speedup
View Details →
View Details →
US Wildfire Trend Analysis — Poisson regression + MLP classification on 1.88M records (1992–2015).
View Details →
View Details →
Data Engineering & Applications
SVD Image Compression App — PyQt6 GUI application with real-time preview, Eckart-Young theorem verification. PSNR 44.7 dB
View Details →
View Details →
Wardrobe Recommendation Database — 15-table relational schema with weather/occasion-aware rule-based recommendation engine.
View Details →
View Details →
GIS Web Mapping & Spatial Analysis — GeoPandas, Shapely, ArcGIS Python API for world cities spatial analysis.
View Details →
View Details →
Ongoing Research
Flight Delay Prediction — M.S. Thesis — Selective prediction framework for American Airlines using XGBoost on 1.1M+ flights (Snowflake). AUC 0.6863 (no ETD) / 0.9280 (+ ETD); AURC 0.2908. Investigating abstention policies to bound risk.
Data under NDA — results only
Data under NDA — results only
Adaptive Learning Platform Analysis — Taichung Education Bureau — Evaluating the effectiveness of an adaptive learning platform on Grade 5 math (127 classes, 2,414 records). Spearman ρ = 0.096 (usage vs outcome); within-school adjusted ρ = −0.132.
Data under NDA — results only
Data under NDA — results only
Experience
2022 – 2024
Bilingual Teaching Assistant
National Chung Hsing University
- Calculus — One-on-one support for international students; led recitation sessions linking calculus theory to numerical algorithms
- Database Systems Design — Assisted with SQL assignments; introduced Power Query; provided bilingual translations of lecture notes
- Quantum Computing — Guided students through Python & Qiskit assignments; explained probabilistic reasoning and simulation tasks
2023 – 2024
Undergraduate Thesis: Quantum Bayesian Inference
Department of Applied Mathematics, NCHU
- Investigated whether quantum computer outputs conform to Bayesian probability laws using Qiskit on IBMQ hardware
- Applied Bayesian updating to analyze qubit measurement outcomes; revealed divergence between actual quantum runs and simulator predictions
2020 – 2024
Part-Time Roles
Various, Taiwan
- Bilingual Conference Staff (NCHU) — Escorted visiting scholars, coordinated logistics, provided real-time bilingual assistance
- Cram-School Instructor & Private Tutor — Mathematics, physics, chemistry, biology, Chinese, and English
- 3D Printing Studio — FDM printer operation, slicing and print workflow management
International Volunteer
- Taught English and art at rural elementary schools; participated in desert-area afforestation projects
Last updated: February 2025 · Save as PDF