Anton Shvets Email & Phone Number

Cupertino, California, United States

Cupertino, California, United States

Beaverton, Oregon

- Developed statistical forecasts and predictive models to estimate unconstrained demand, and allocation for limited edition footwear launching on the SNKRS App across the multi-geo marketplace.- Reduced demand forecasting error by 50%+ globally through improved forecasting and predictive modeling- Oversaw research and algorithm development across the product team; pulled and preprocessed data, and built ML pipelines in production supporting a WebApp & API, and provided automated hindsight accuracy reporting.- Set up experiments for hypothesis testing to better understand consumer trends and behaviors, identifying and aligning new metrics to reach business objectivesTools/Languages: Python, SQL, Tableau, Snowflake, S3, EMR, Athena, Pandas, Numpy, SciPy, Scikit Learn, Tensorflow, Keras, Github, Plotly

Beaverton, Oregon

- Collaborated on projects across demand sensing, size allocation, product recommendation, and forecasted app growth for the SNKRS app.

- Performed analysis of the home energy consumption data for the ChargeForward program.- Predicted driving and charging behavior for households that are participating in smart charging in order to reduce their consumption cost by optimizing charging times.- Clustered groups within the pilot program with similar behavioral characteristics.Tools/Languages: Python, R, SQL, S3, EMR, Athena, Pandas, Numpy, SciPy, Scikit Learn, Github

San Francisco Bay Area

All my projects can be found on my github repo: https://github.com/shvetsanton/NBA Analysis - Advanced Statistics (Bayes Statistics) Project- Built visualizations to demonstrate several progressions in the league since the 1980's.- Draft pick optimizer and mid-season identifier of good players (empirical bayes model). Natural Language Processing Project — Quora Question Pairs:- Predict if a pair of questions are duplicates using word embedding and Deep Learning.- Compute the Word Movers Distance using Word2Vec embeddings, and use logistic regression to predict duplicity probability. - Deep Learning approach: Sequence2Sequence model with LSTM encoder & decoder. Deep Learning Project — Understanding the Amazon From Space:- Built a multi-class classification algorithm to label satellite image chips with atmospheric conditions and various land cover and use. - Helps the global community better understand where, how, and why deforestation happens all over the world, and how to respond.Weather Streaming Project:- Continuously stream weather data using a python script on an EC2 instance and S3 for storage.- Clean and structured data was then stored in a Postgres Database. - Train SVC model to predict city of origin to demonstrate the different weather patterns of the observed cities. Model was optimized via a nested cross-validation for optimal hyperparameters.Star Cluster Simulation Project:- Performed exploratory data analysis and statistical analysis on a globular cluster. - Fit various distributions for the radius and the velocities of the stars at different time shots during the simulation. Used the distributions to approximate the velocity dispersion and the effective radius of the cluster.Tools/Languages: Python, S3, EMR, Pandas, Numpy, SciPy, Scikit Learn, Tensorflow, Keras, Github, Jenkins, Circleci

Santa Barbara, California Area

I was part of a research team at UCSB studying cold and hot gas behavior in the intergalactic medium. I was responsible for removing cosmic rays in images using a python script that utilized a Laplacian kernel. These images were used as a base for this study.Our team would have weekly meetings in which we discussed our progress in our individual assignments as well as recent studies done in our field.

Master Of Science (M.S.), Data Science

Bachelor’S Degree, Physics, Minor: Astronomy

Education record

Activities and Societies: Alpha Gamma Sigma Honor Society, Toastmasters