SAM BELLOWS
DATA SCIENCE MACHINE LEARNING NLP EXPERT
CONTACT
adczcq@r.postjobfree.com
New York, NY
linkedin.com/in/SBellowsDataScience
github.com/sbellows1
PROFILE
Data Scientist who specializes in
building recommender systems and NLP
models.
EDUCATION
Expected December 2020
CUNY School of Professional Studies
Master of Data Science
2015
New York University
Bachelor of Fine Arts
TECHNICAL SKILLS
Python, R, SQL
Machine Learning
Natural Language Processing
Collaborative Filtering and Matrix
Factorization
Statistics and Probability
Data Visualization
Calculus and Linear Algebra
Data Cleaning and Wrangling
Git, Excel, Tableau
SOFT SKILLS
Communication
Collaboration
Critical Thinking
Problem-Solving
Attention to Detail
Team Player
EXPERIENCE
2019 - PRESENT
Independent Data Scientist
NLP Models
Developed a sentiment analysis classifier using Naïve Bayes to improve AUC by 40% over the baseline model. The model would allow companies to quickly find and address negative feedback to improve customer experience and customer retention.
Created a highly scalable RNN sentiment classifier using Keras that increased F1 score by 28% over the baseline model and significantly outperformed all traditional machine learning models. MovieLens Recommender System
Created a collaborative filtering recommendation engine using a FunkSVD algorithm that improved recommendation RMSE by 23.5%. This could potentially lead to large increases in customer satisfaction or an increase in sales depending on the use case.
Used statistics to create model basis and included hyperparameter optimization and regularization.
Salary Prediction Model
Created an end to end machine learning pipeline including an ETL pipeline, feature engineering, multiple machine learning models, and hyperparameter optimization that resulted in a 47.5% decrease in model error compared to the baseline model.
Employed Object Oriented Programming and other best coding practices for readable, reusable, and professional code.
Saved results to a SQL database for ease of access. 2015–Present
Classroom Instructor A-List Education
Designed and taught SAT classes by investing in each individual student and providing personalized feedback and goals.
Continuously improved my instruction and student results by incorporating both mentor and student feedback.
Achieved an average score increase of 120 points per student on the SAT, a 71% increase over the national average score increase.
Stayed organized and prioritized effectively, helping me teach 15-20 classes per year, or 300 – 400 students.