* Owned all product development with direct P&L impact. Led strategic pricing, inventory, promotions, and retention efforts. Managed product and applied science in the revenue org.* P&L responsibility for ThredUp’s core U.S. business ($XXXM)

* Responsible for pricing, inventory, and promotion strategies; responsible for P&L of core U.S. business.* Delivered multi-armed bandit system with automated reporting. Lower friction drove an order of magnitude increase in experimentation cadence. MAB-driven price adjustments during promotional events increased revenue by 1.5%.* Developed shared view of inventory and shared metrics used across the company. Used improved listings to drive 15% year-over-year growth in monthly active buyers, with 4% less marketing spend and 50% less shoppable inventory. Expanded contribution margin 500 bps in an environment with rising ops and labor costs.* Developed and coordinated plan to successfully close a $2.5M revenue gap in the final 30 days of Q3’2023. Efforts spanned FP&A, product, operations, supply, growth marketing, & promotion teams.

* Responsible for a the portfolio of internal products used to manage ThredUP's two-sided marketplace. Managed two cross-functional teams.* Developed & executed on roadmap to introduce user-level randomized controlled trials for pricing. Used this system to demonstrate that we can measure elasticity in a single-SKU marketplace & unlocked 10% revenue growth through pricing optimization.* Oversaw development of machine-learning based pricing, which simultaneously increased both customer growth and profitability. * Worked with ThredUp’s legal team to draw clear boundaries around ethical A/B testing practices and use of machine learning systems.

* Managed a team that improved ThredUP's operational efficiency through automation and experimentation.* Introduced the first A/B testing framework for operations and used it to conduct experiments that are now saving the business millions of dollars per year.* Led development of machine-learning based anomaly detection system, allowing ThredUp to reduce errors with 75% fewer staff. Developed batching concept for QA later demonstrated to boost operator speed by a factor of 10.* Worked across ThredUp operations to build a bottom-up, data-driven culture.

* Conceived of and built thredUP's first deep learning system, automating image tagging work previously done by hourly staff while improving accuracy* Built system to orchestrate inbound and outbound conveyors in network of distribution centers* Led development of platform for customer service tools* Introduced our company to React; created onboarding guide to teach best practices to new hires* Mentored other engineers

* Developed systems and introduced data governance policies that reduced the cost of ThredUp’s petabyte-scale photo storage needs by over 90% while maintaining performance and quality.* Built a photo post processing service that lifted the quality of pictures across ThredUP's customer facing website while making the application stack more resilient.

Developed and implemented crowdsourcing strategy to train first commercial Watson supercomputer on a new domain - Devised A/B tests to more than double percentage of useable crowdsourced data - Sped up critical path item by months while driving down costs for training question and answer sets by over 80%Created command line tools in Ruby and Python used for data ingestion and building Watson’s language modelResearch on hydrogen as a lifting gas and statistical analysis of tethered blimp use in Afghanistan and Iraq, leading the Army to further pursue hydrogen as an option

Developed methodologies to assess effectiveness of Intelligence, Surveillance, and Reconnaissance (ISR) capabilities in CT operations in Iraq and Afghanistan- Crunched tens of thousands of records using Excel, Python, and other desktop tools- Results led to immediate reallocation of over $2B to ISR by Secretary of Defense and Joint Chiefs of StaffCreated data storage and management framework from scratch using Ruby and MySQL to clean, store, and analyze millions of intelligence records for multiple clients- Used to deliver the first quantitative assessments of sharing in the Intelligence Community, enabling leadership to gain better access to reportingLed MQ-9 Reaper acquisition guidance for the United States Air Force- Leveraged years of studies and operations research to make recommendations on sensor mix and platform count across a range of units and missions in less than three months- Uncovered incorrect assumptions in USAF’s then-current acquisition strategy, leading clients to change their mental model of platforms required to meet mission requirements