CV

A PDF version is available on request — reach out at cyf0906@stanford.edu.

Education

• M.S. in Data Science, Johns Hopkins University, Jan 2026 – Dec 2026 (expected)
  • GPA: 4.0 / 4.0
• M.A. in German Studies — Specialization in Computational NLP & Textual Data Science, Stanford University, Sep 2024 – Jun 2026 (expected)
  • GPA: 3.6 / 4.0
  • Research focus: NLP, topic modeling, and multimodal analysis applied to historical German textual and visual corpora.
• B.A. in German Studies, University of California, Santa Barbara, Sep 2019 – Sep 2023
  • GPA: 3.74 / 4.0
  • Related coursework: Applied Mathematics, Data Science

Skills

• Programming Languages
  • Python, R, SQL, Java, HTML & CSS
• Machine Learning & Statistics
  • Tree-based models (Random Forest, Gradient Boosting, LightGBM)
  • Regression models (Linear, Logistic, regularized)
  • Support Vector Machines
  • Unsupervised methods (k-means, PCA, BERTopic, LDA)
  • Model evaluation, cross-validation, hyperparameter tuning
  • Feature engineering, MLOps
• Data Science & Visualization
  • pandas, NumPy, SciPy, scikit-learn
  • seaborn, Matplotlib, Plotly, ggplot2 (R)
  • Apache Parquet
  • R Markdown, Jupyter
• Cloud & Tools
  • AWS (S3, SageMaker AI, EC2)
  • Google Cloud BigQuery
  • IBM Watson Studio
  • GitHub, Flask, Tableau
• Specialties
  • Natural Language Processing (spaCy, BERTopic, MALLET)
  • Multimodal analysis (vision-language models, OCR pipelines)
  • Fraud detection and risk modeling
  • Class-imbalance methods (resampling, reweighting, threshold tuning)
• Languages
  • Mandarin Chinese — Native
  • English — Fluent (US High School through Masters)
  • German — Fluent

Certifications

• IBM Data Science Professional Certificate — Fall 2025
• Google Data Analytics Professional Certificate — Summer 2025

Selected Projects

A curated list of data science, machine learning, and computational humanities projects. Full writeups, GitHub links, and additional detail live on the Projects page.

• Mosaik Reverse Prompting + Multimodal Analysis (ongoing, Stanford)
  • Multimodal pipeline (YOLO panel segmentation + OCR + VLM/LLM reverse prompting) processing 26K–64K mid-20th-century German comic panels into structured semantic representations.
• Credit Tier Prediction ML Pipeline (Volunteer Data Scientist, Microsoft DS Team, Summer 2025)
  • End-to-end pipeline on 4M+ financial records. Improved accuracy 4% and reduced false positives 6% via ensemble modeling and feature engineering.
• Transaction Fraud Detection ML Pipeline (Volunteer Data Scientist, Google, Winter 2023)
  • Production-ready GBDT model on 800K+ Google Pay transactions. Reduced false positives 8.7% and improved recall 2.87%.
• E-Commerce Profitability Analytics — Amazon India (Spring 2025 – Oct 2025)
  • R Markdown project on 1.2M+ transactions: clustering, regression, and tree-based models for profitability drivers and pricing optimization. GitHub
• German Romanticism Corpus Analysis (1780–1840) (Sep 2025 – Dec 2025, Stanford)
  • Custom corpus + NLP pipeline (spaCy, BERTopic, MALLET, R tidytext) tracking thematic evolution across early, transitional, and late Romanticism.
• Credit Fraud Detection (Mar 2025 – Jun 2025)
  • Class-imbalance study (500 frauds in 200K+ records) benchmarking Random Forest, SGDClassifier, and MLP with PCA, downsampling, and reweighting.
• SpaceX Falcon 9 Launch Analysis (Nov 2025)
  • EDA, ML models (Logistic Regression, Decision Tree, SVM), and interactive Plotly/Dash dashboards on Falcon 9 launch data. GitHub
• Gomoku AI Engine — Minimax + RL (Sep 2024 – Mar 2025, Stanford)
  • Two-phase project: Minimax with Alpha-Beta pruning baseline, then self-play and Q-learning extension. GitHub
• Psychological Impact Prediction (Jun 2024 – Sep 2024)
  • LightGBM classifier for psychological risk prediction on imbalanced healthcare data. AUC 0.87, with interpretable visualizations for clinicians.

Contact

• Email: cyf0906@stanford.edu
• Phone: 626-536-6862
• LinkedIn: peter-cheng-a2436a331
• GitHub: PeterCheng0906