Career Profile
Data Scientist with 3+ years of experience. Adept at understanding business problems, applying supervised/ unsupervised machine learning, and presenting to technical and non-technical audiences. Have industry experience in Natural Language Processing, Data Visualization, and Optimization. Currently, on a team at Cox Automotive building data solutions for product teams.
Work Experience
Cox Automotive is a global automotive business that includes Kelley Blue Book, Xtime, Autotrader.com, and Manheim. Currently, working as a Data Scientist building web applications, deploying APIs, and applying machine learning to solve problems for product teams at Cox.
UTC is an American multinational conglomerate. Our team collaborated with the human resource (HR) analytics team led by Pankaj Prakash (Director, Data Science) on a system to track and classify HR requests from internal employees to reduce the workload of the HR department.
- Preprocessed 350k unstructured text requests and extracted features for model building using Word2Vec.
- Implemented a topic modeling pipeline to categorize incoming HR questions from employees.
- Designed visual and diagnostic Tableau dashboards to track model classifications and trending HR questions.
Urban Sharing is a Norwegian tech-startup that has developed a technology platform for forward-thinking mobility solutions. Worked on the data science team led by Hans Martin Espegren on developing more intelligent ways to rebalance their bike share systems.
- Coded algorithms to calculate the optimal hourly distribution of bikes across the combined 500 station system.
- Developed statistical metrics to incentivize van drivers to rebalance bikes to stations to increase user trips.
- Presented research and progress with a live demo to the CEO, software, and hardware teams (~40 people).
- Created a real-time map to highlight stations that should be prioritized to aid van drivers in rebalancing efforts.
Foresight Associates is a management consultancy led by Vittorio Raimondi. Was part of a small team providing consulting services for various marketing and insight teams of The Coca-Cola Company.
- Performed regression analysis and ad-hoc analysis of consumption data to determine brand affinity.
- Structured drink consumption data to be fed into system dynamic models to predict brand churn.
- Managed file systems to ensure that Infotools drink surveys were up to date for 50+ global Coca-Cola markets.
- Delivered PowerPoint decks to clients that conveyed a clear story about key drivers of growth/loss for brands.
Worked under Matthew Richey (Professor of Mathematics) on applying statistical learning algorithms on St. Olaf’s admissions data to make better enrollment decisions.
- Analyzed four years of acceptance and student interaction data to feed into enrollment prediction models.
- Explored the price sensitivity of merit/financial aid recipients and documented where further data is needed.
Projects & Competitions
155th / 2227 ~ Top 7% in Kaggle competition, which involved fine-tuning state of the art NLP models (ex. BERT, RoBERTa) using Huggingface’s PyTorch library to isolate the portion of the tweet that reflected the labeled sentiment.
MVU is a non-linear dimensionality reduction algorithm. Delivered a report and a 50-minute class presentation on the mathematical derivation and pros/cons of this unsupervised learning algorithm when tested on toy datasets.
Designed an innovative visualization to inspect text misclassifications within a confusion matrix built in Tableau.
Wrote a version from scratch in Numpy and implemented a couple of GAN variations in Pytorch. Trained neural networks on GPUs using computing clusters with bash scripts in a terminal.
Skills & Proficiency
Code
Python
R & RStudio
Pytorch & Keras
Javascript & Node.js
HTML & CSS
Database
MySQL
Google Cloud & BigQuery
AWS
Visualization
ggplot2
Tableau
D3.js
Research Experience
Collaborated with a group of professors and students on a National Science Foundation funded project regarding disrupting human trafficking networks. Tasked with exploratory data analysis of international border control interception data to identify important variables that could be leveraged to inform operations research models.
Provided statistical analysis (T-test & ANOVA) for a faculty research project concerning the effects of alcohol on adult and adolescent mice. Created multiple view visualizations (ggplot2) to inspect the variance across experiment factors. Presented our research at the National Conference for Undergraduate Research in Memphis, TN.