Biography

Russell Allgor is the chief scientist for Amazon.com, where he leads a team of mathematical modeling experts. Allgor has demonstrated outstanding accomplishments in practice, management and research of OR/MS at Amazon. He is one of the most influential scientists in the field of logistics and fulfillment systems for e-commerce. He and his team focus on using data analysis, modeling, simulation and optimization methods to improve the efficiency of Amazon’s operations. He has focused on problems including network design and facility location, inventory planning, order assignment, equipment and process design, vehicle routing, and process control within and across facilities. Ideas and algorithms developed by Allgor and his team have returned billions of dollars to Amazon’s bottom line. Prior to joining Amazon.com, Allgor worked in applied R&D for Bayer AG in Leverkusen, Germany. He has a PhD in chemical engineering from MIT and a Bachelor of Science from Princeton University. He is a member of the National Academy of Engineering and an INFORMS fellow.

Abstract

Amazon has continued to grow with sales from Amazon retail and Fulfillment by Amazon sellers, requiring expansion of the fulfillment network and insourcing most of the outbound transportation. However, this growth and the shift from a 3P toward a 1P transportation network uncovered challenges of increased complexity and that lacked economies of scale. Even with automated tools, the network was difficult to manage and costs were increasing. To continue to grow the business, we needed to redesign the network to reduce cost, improve speed and generate economies of scale. We transitioned our design approach from optimizing 3P rate cards to optimizing purchased assets and their corresponding utilization, so that we can continue to raise the bar by offering superior customer experience at the fastest speed, broadest selection and lowest prices. Like most big ideas at Amazon, we focused on inventing and simplifying the design. We created a structure for fulfillment that matches capacity with customer demand in regional clusters, allowing transportation to scale, reducing controllable variability and expecting inventory placement to follow.