I'm a Master's student in Statistics and Data Science at Cornell University with a passion for building data-driven systems that are as impactful as they are intelligent. I thrive at the intersection of analytical rigor and real-world application, and I’m currently seeking full-time data scientist, data analyst, data engineer opportunities where I can turn data into decisions that matter.
- MPS in Statistics and Data Science | Cornell University
- Bachelor of Arts in Math & Applied Psychology | Boston College
- Programming/Big Data: Python (Pandas, Matplotlib, Pytorch), Hadoop, Hive, Java, SQL, R, Excel, SAS, SPSS, Tableau, AWS
- Machine Learning: Classification, A/B Testing, NLP, Language Modeling, Generative Models, CNN, KNN, Regression
An interactive cocktail recommendation tool powered by TheCocktailDB API. Users can explore drinks based on ingredients, compare options via A/B testing, and receive personalized suggestions based on their preferences.
- Features: Multi-tab UI, image-based drink results, A/B preference tracking, personalized drink suggestions
- Visualization: Dynamic charts for user preference trends (glass type, alcohol, ingredients)
- Tech Used:
ipywidgets,pandas,requests,matplotlib,seaborn,Jupyter
An NLP project for classifying emotional tone in text using machine learning. Developed as part of CS 3780/5780, this project compares traditional CNN + Bag-of-Words with the transformer-based DistilBERT model.
- Best Model: DistilBERT (accuracy 0.77, above baseline)
- Techniques: Supervised learning, feature extraction, transformer embeddings
- Tech Used:
pandas,PyTorch,Transformers,Jupyter
A graduate group project from STSCI 5954 that analyzes data from 777 U.S. colleges and universities. Using exploratory data analysis and machine learning models (Logistic Regression and Random Forest), the project classifies institutions as elite or non-elite based on performance metrics like application rates, graduation rates, and alumni donations.
- Best Model: Random Forest (highest accuracy and ROC AUC)
- Key Insights: Importance of out-of-state tuition, academic reputation, and student services
- Tech Used:
pandas,scikit-learn,matplotlib,seaborn,Jupyter
Ithaca, NY · Jan 2025 – Present
- Developed a scalable pipeline for 15K+ time-series rows using TINGO and FRED APIs; optimized preprocessing via batch ingestion, feature engineering, and imputation — reducing data prep time by 30%
- Integrated PyTorch training with AWS SageMaker, leveraging GPU acceleration to handle 396-feature datasets efficiently and enable retraining for forecasting cycles
- Trained a Temporal Fusion Transformer on financial and macroeconomic data, boosting stock prediction accuracy by 15% and improving investor-style alignment by 38% without compromising precision
Boston, MA · Feb 2024 – May 2024
- Created and deployed the Gender IZU (Initiative to Zero Violence) survey using REDCap and Excel (VBA), collecting data from 1,000+ families to evaluate domestic violence interventions
- Built and normalized multi-department relational databases in SQL Server, improving query speeds from 20 minutes to under 1 second via optimized joins
- Validated survey logic and pipelines, improving data integrity and reducing entry errors by 40%, enhancing public health insights and collaboration
Boston, MA · Sep 2023 – Dec 2023
- Designed and maintained relational databases for 50+ banks using Microsoft SQL Server, improving access to 200+ tables and standardizing financial data reporting
- Designed KPI-aligned pre/post surveys on budgeting confidence; visualized 20% post-rollout improvement using Tableau
- Cleaned and analyzed 400+ bank records with Excel (VBA, PivotTables) and R (ggplot2); identified low account-opening rates among youth and proposed outreach to five new schools
Boston, MA · Nov 2022 – Sep 2024
- Preprocessed student performance and demographic data from 4,000+ schools across 20+ counties using SPSS, boosting efficiency in Apple classroom product deployment by 30%
- Created 5+ interactive Tableau dashboards for Apple HQ and faculty teams, increasing the adoption of teacher-centric features by 20%
- Conducted correlation analysis between AI-coded and manually coded qualitative data to identify the top 5 barriers to device adoption, presented at Apple Education’s annual conference
- Email: tw584@cornell.edu