Quantitative Analyst, Machine learning, Pyton, R, SQL, Pytorch,Tableau

Stephanie Wei

*************@*****.*** +1-984-***-**** 2840 Jackson Ave., Queens, NY 11101

EDUCATION

Duke University, Fuqua School of Business North Carolina, US Master of Quantitative Methods and Management Jul 2023 - May 2024 New York University, College of Arts and Science New York, US Bachelor of Mathematics Aug 2019 - May 2023

SKILLS

Programming Language: Python, Java, JavaScript, SQL, R Packages: Scikit-Learn, Pytorch, TensorFlow, NumPy, Pandas, Matplotlib, BeautifulSoup, NLTK, SciPy, tidyverse ML Models: LSTM, Attention, CNN, Statistical Analysis, Linear/Logistic Regression, xgboost-gpu, random forest WORK EXPERIENCE

MirWork New York, US

Quantitative Analyst Intern Jul 2024 – Present

Designed and implemented a scalable ML pipeline for data preprocessing and model training. Integrated risk assessment checkpoints to ensure model reliability and compliance with regulatory requirements.

Assessed model risk by performing comprehensive evaluations of model performance under various stress scenarios, documenting findings in detailed model risk reports to meet internal and external audit standards.

Leveraged Pandas and SQL to cleanse and normalize interviewee records and interview session data. Conducted K- Means clustering to identify key risk-related features such as skill gaps, session trends, and engagement metrics.

Implemented Random Forest and XGBoost algorithms to predict user churn rate, integrating explainability methods

(e.g., SHAP values) to ensure interpretability and compliance with risk governance policies. Optimized the model to a 95% accuracy and 82% recall rate.

Developed a framework to identify and mitigate potential biases in ML models, conducting ongoing performance reviews to minimize operational and reputational risks associated with inaccurate predictions.

Segmented users into high-, medium-, and low-risk categories based on churn probability, enabling the development of targeted strategies to mitigate financial exposure and improve resource allocation efficiency. ISTARI Venture New York, US

Quantitative Analyst Intern Mar 2024 – June 2024

Wrote and implemented web scraping scripts using Python and bs4 to collect market and operation data from 10+ crypto company websites, and conducted exploratory data analysis (EDA) to define project scope.

Utilized LSTM for time-series risk forecasting incorporates both historical operational data and policy data with Tensorflow, achieving model performance of 0.221, 2.6% measured in RMSE and MAPE.

Optimized the deep learning model through implementing Attention Mechanism to the LSTM based model to represent the weight of each time period of the collected company market and operational data.

Reduced model training time by 20% by applying PCA to preprocess and reduce dimensionality.

Developed 10+ interactive company risk assessment dashboards using Power BI, integrating 15+ key financial and operational risk metrics, empowering data-driven risk mitigation strategies for investment decision-making. PROJECTS

Citi Market Risk & Portfolio Management Nov 2024

Developed and implemented a comprehensive DCF model in Python to evaluate the company’s enterprise value, performing scenario analysis and sensitivity tests.

Completed 50 pages of detailed documentation to explain model assumptions, calculations, and methodologies.

Constructed a diversified stock portfolio of 20+ equities across multiple sectors, optimizing for risk-adjusted returns using a combination of fundamental and technical analysis.

Evaluated portfolio performance using advanced metrics, including calculating alpha, beta, and the Sharpe ratio, providing insights into portfolio risk and return characteristics. Duke House Price Prediction Competition, First Prize Nov 2023

Extracted house price insights from Denver datasets, including housing area and distance data with SQL, and enhanced query performance and data retrieval efficiency through indexing and batch processing techniques.

Conducted data visualization using Seaborn to create correlation matrix heatmaps and histograms, leveraging representative samples to optimize performance and efficiency.

Built a multilayer perceptron network(MLP) to examine more than 10,000+ data entries, achieving an accuracy rate of 92.5% through hyperparameter tuning and feature engineering in Python.

Contact this candidate