I led the machine learning efforts driving DoorDash's growing New Verticals offering (grocery, convenience stores, and more retail) helping us expand into more and more areas of retail delivery. We built production ML systems to solve a range of challenging problems spanning search & recommendations, inventory management, and fulfillment/logistics.

I led the team developing the core recommendation algorithms that allow Stitch Fix to delight clients with new apparel that fits (or expands!) their personal style. Our recommendations played a central role in nearly all of the company's sales. We constantly experimented with new algorithms, data sources, and experimentation methods in a field where delivering personalized recommendations is both challenging and high impact.

South Park Commons is a learning community in which members experiment with ways to improve the world with technology.

I took a sabbatical from a full time role to explore applications of data science and AI in different fields through consulting projects and volunteering with political and non-profit organizations.

Led teams working on efforts to diagnose and improve performance of our core home valuation algorithm.

I joined Color at an early stage as the first data scientist and established our analytics and reporting framework, helping the company become more data-driven across the board by partnering with a range of teams including product, laboratory, genetics, and sales. I built out our data science team to scale with the company's growth.

I led a diverse, talented team of data scientists who worked closely with engineering and product teams to keep making Twitter better for our users. Our work encompassed projects in machine learning, experimentation, and analysis to guide product strategy.My statistical interests span a range of topics in prediction, clustering, multilevel models, time series forecasting, and LDA and related probabilistic topic models. Areas of application include understanding user behavior and interests, international growth and comparative "macro"-level analyses, and the influence of media on Twitter.