Zachary Owen, engineer and researcher, working on applications of machine learning and optimization.
I presented my thesis research on dynamic pricing and assortment optimization at the MSOM 2018 conference in Dallas, TX.
I presented new updates to my thesis research at the INFORMS 2018 Annual Meeting in Phoenix, AZ.
My work at Armoire and the company as a whole was recently profiled in GeekWire, Seattle's premier source for tech news.
Education
Massachusetts Institute of Technology
Ph.D. in Operations Research
Advisor: Prof. David Simchi-Levi
Thesis Committee: Stephen C. Graves, John N. Tsitsiklis
Thesis: Revenue Management and Learning in Systems of Reusable Resources
In my thesis research I used linear programming to derive policies with provable constant-factor performance guarantees for managing pricing and assortment strategies for large-scale systems of reusable resources. The resulting policies are applicable to settings such cloud computing, city-scale parking management, and clothing rentals.
Cornell University, College of Engineering
B.S. in Engineering
Major: Operations Research and Information Engineering
Minor: Applied Mathematics
Graduated magna cum laude
Conducted research in statistical learning with Prof. Peter Frazier, specifically in adaptive policies for stochastic non-convex optimization such as stochastic bisection search to optimize an unknown functions subject to measurement error. These problems were inspired by the challenge of optimizing dosages in medical applications.
Experience & Research
Co-Founder, Machine Learning & Engineering
Armoire Style
Responsible for the development and maintenance of all technical systems at Armoire including the primary web application, internal applications, and machine learning systems used in style and fit recommendation. Designed and implemented in production custom machine learning algorithms, based on user-item embeddings, for contextual clothing recommendation and fit prediction.
Research Assistant
Massachusetts Institute of Technology
Price and Assortment Optimization for Reusable Resources (R&R, Management Science, 2018)
Z. Owen and D. Simchi-Levi
Developed a constant factor performance guarantee for a large-scale pricing and assortment optimization of reusable resources under a continuous time horizon.
A Statistical Learning Approach to Personalization in Revenue Management (under revision, 2018)
Chen, X., Z. Owen, C. Pixton, and D. Simchi-Levi
Derived finite-sample performance bounds for MLE-based policies in the context for pricing and assortment optimization.
Selected Talks: with peer-reviewed extended abstracts
- MSOM Annual Conference. Dallas, TX. July 2018.
Revenue Management in Systems of Reusable Resources with Random Service Times and Time-varying Demand - INFORMS Revenue Management and Pricing Conference. Amsterdam, NL. June 2017.
Price and Assortment Optimization for Reusable Resources - MSOM Annual Conference. Toronto, ON. June 2015.
A Statistical Learning Approach to Personalization in Revenue Management - MSOM Annual Conference. Seattle, WA. June 2014.
Price Differentiation: A Machine Learning Approach
Lead Data Scientist
Hive Maritime
Developed algorithms for prediction of destination and time of arrival for the global shipping fleet based on more than 300GB of satellite AIS data encompassing hourly location data for every seagoing vessel on the planet for over a year. Used nearest-neighbor style algorithm on GPS trajectories to predict the destination and ultimate time of arrival for transpacific voyages within hours of departure. Predictions on held out data were generally within hours of true arrival time.
Data Science Intern
Stitch Fix
Developed a strategy for measuring inventory lifetime value based on the expected number of shipments and the opportunity cost each shipment selection. Applied this metric to improve inventory health through data-driven clearance strategies. Took the project from the idea stage through to a dashboard used by decision makers in the merchandising team.
Trading Analyst
Structured Equity Derivatives
Barclays Capital
Responsible for pricing, trading, and hedging equity-linked structured notes in both primary and secondary markets including reverse convertibles, auto-callables, lookback notes, etc. Worked with senior traders to manage the risk inherent in a book of exotic equity derivatives including sensitivities to implied volatility, interest rates, market gaps, and higher order risks.
Technical Skills
Programming Languages
Python, Django Framework (proficient)
Julia, R (prior experience)
Data Analysis
SQL (Postgres/PostGIS), NumPy, Pandas
Machine Learning
Scikit-learn, PyTorch
General
Git, AWS (EC2, S3, RDS, etc.), Gurobi, Algorithms & Data Structures
Interests
- Machine learning and its applications in various domains
- Probabilistic Modeling
- Recommender Systems
- Natural Language Processing
- Euclidean and Non-Euclidean Embeddings
- Time Series Modeling
- Reinforcment Learning
- Designing and building intelligent systems from theory-driven prototypes to production implementations.
- Design and analysis of data-driven algorithms
- Optimization of stochastic systems
Relevant Courses
- Inference & Information
- Machine Learning
- Statistical Learning Theory
- Introduction to Mathematical Programming
- Nonlinear Programming
- Introduction to Algorithms
- Structure of Information Networks
- Special Topics in Applied Statistics (Time Series)
- Statistics for Financial Engineering
- Stochastic Calculus for Financial Engineering II
Contact
610-401-1035
|
zachdowen@gmail.com
Tortilla aficionado. Avid bike commuter and long-time Steelers fan. Cat person. Currently based in Seattle, WA.