Overview
Held each fall, the annual meeting brings together over 6,500 people to the world's largest operations research and analytics conference. It features more than 800 sessions and presentations, opportunities to meet with leading companies, universities and other exhibitors, an onsite career fair connecting top talent with over 100 organizations at the forefront of O.R. and analytics application, and other networking and educational events. The theme for the INFORMS Annual Meeting this year is “Smarter Decisions for a Better World” which encapsulates the core ethos of the organization and the mission it strives to achieve.
Amazon Science is a diamond sponsor of the INFORMS Annual Meeting this year and Fulfillment by Amazon (FBA) is a platinum sponsor of the INFORMS forum for Women in OR/MS (WORMS) this year.
Sponsorship Details
Amazon Organizing Committee
-
General chair -
Prize committee
-
Roundtable member
-
Arrangements co-chair
-
INFORMS Fellow, 2024
-
INFORMS 2025 President Elect
Sessions
LandMover Equipment Company: Optimizing Servicizing Product Contract
October 20, 8:00 AM - 9:15 AM PDT
Website: Link
Case Competition:
Presented by Saurabh Bansal, Dan Guide, Wei Wu, Yinshi Gao, and Yinshi (Alice) Gao
Room: Regency - 605 / Snohomish
Case Competition:
Presented by Saurabh Bansal, Dan Guide, Wei Wu, Yinshi Gao, and Yinshi (Alice) Gao
Room: Regency - 605 / Snohomish
Transportation Marketplace Rate Forecast Using Signature Transform
October 20, 9:45 AM - 2:00 PM PDT
Website: Link
Daniel H. Wagner Competition II:
Presented by Haotian Gu, Xin Guo, Tim Jacobs, Philip Kaminsky, and Xinyu Li
Room: Regency - 605 / Snohomish
Daniel H. Wagner Competition II:
Presented by Haotian Gu, Xin Guo, Tim Jacobs, Philip Kaminsky, and Xinyu Li
Room: Regency - 605 / Snohomish
Design of Experiments: From A/B Testing to Adaptive Experimentation
October 20, 10:45 AM - 12:00 PM PDT
Website: Link
Presentation: "Adaptive Experimentation Methods at Amazon", 11-11:15am, presented by Tanner Fiez
Room: Summit - 435
Presentation: "Adaptive Experimentation Methods at Amazon", 11-11:15am, presented by Tanner Fiez
Room: Summit - 435
Optimizing Last Mile Efficiency at Amazon
October 20, 8:00 AM - 9:15 AM PDT
Website: Link
Presentations:
"GPU Based Route Planning with Cuopt", 8 - 8:18am, presented by Bhargav Kunkulagunta
"Energy Aware Last-Mile Route Planning", 8:18 - 8:36am, Andre Snoeck, Aniruddha Bhargava, Daniel Merchan, and Mahmood Zangui
"Route Time Estimate Optimization with Non-Linear Opportunity Cost Awareness", 8:36 - 8:54am, presented by Ivan Tereshchenko
"Handling Multi-Objective Contiguous Unit Allocation Problems Through Resources in a Graph Modeling", 8:54 - 9:12am, presented by Julie Poullet and Andre Snoeck
Room: Summit - Ballroom 2
Presentations:
"GPU Based Route Planning with Cuopt", 8 - 8:18am, presented by Bhargav Kunkulagunta
"Energy Aware Last-Mile Route Planning", 8:18 - 8:36am, Andre Snoeck, Aniruddha Bhargava, Daniel Merchan, and Mahmood Zangui
"Route Time Estimate Optimization with Non-Linear Opportunity Cost Awareness", 8:36 - 8:54am, presented by Ivan Tereshchenko
"Handling Multi-Objective Contiguous Unit Allocation Problems Through Resources in a Graph Modeling", 8:54 - 9:12am, presented by Julie Poullet and Andre Snoeck
Room: Summit - Ballroom 2
Order Fulfillment Optimization within 4 walls of Amazon Warehouses
October 20, 10:45 AM - 12:00 PM PDT
Website: Link
Presentations:
"Optimizing Storage Across Bin Types in Traditional Fulfillment Centers", 10:45 - 11am, presented by Kevin Bunn
"Manual Pick Scheduling Optimization in Traditional Fulfillment Centers", 11 - 11:15am, presented by Venkatachalam Avadiappan and Kay Zheng
"Reinventing Robotic-Assisted Picking Algorithms in Amazon", 11:15 - 11:30am, presented by Kay Zheng and Ignacio Erazo
"Prioritizing Demand Picking with Multiple Objectives", 11:30 - 11:45am, presented by Elcin Cetinkaya, Prem Viswanathan, and Kevin Bunn
"Scaling Up Pick Planning in Amazon", 11:45am - 12pm, presented by Ignacio Erazo and Kay Zheng
Room: Summit - Ballroom 2
Presentations:
"Optimizing Storage Across Bin Types in Traditional Fulfillment Centers", 10:45 - 11am, presented by Kevin Bunn
"Manual Pick Scheduling Optimization in Traditional Fulfillment Centers", 11 - 11:15am, presented by Venkatachalam Avadiappan and Kay Zheng
"Reinventing Robotic-Assisted Picking Algorithms in Amazon", 11:15 - 11:30am, presented by Kay Zheng and Ignacio Erazo
"Prioritizing Demand Picking with Multiple Objectives", 11:30 - 11:45am, presented by Elcin Cetinkaya, Prem Viswanathan, and Kevin Bunn
"Scaling Up Pick Planning in Amazon", 11:45am - 12pm, presented by Ignacio Erazo and Kay Zheng
Room: Summit - Ballroom 2
Transportation and Network Optimization at Amazon
October 20, 12:45 PM - 2:00 PM PDT
Website: Link
Presentations:
"Optimizing Amazon's Network Flow Plan: A Decomposition Approach in Balancing Truck Fill Rates and Labor Utilization", 12:45 - 1:03pm, presented by Hyemin Jeon, Theodoros Pantelidis, and Martin Bagaram
"COPPER: Contract Price Optimization Using Willingness-to-Pay Models for Amazon Freight", 1:03 - 1:21pm, presented by Yebin Tao, Juan Xu, and Roger Lederman
"Air Cargo Revenue Management: A Two-Stage Stochastic Program with Learning-based Recourse", 1:21 - 1:39pm, presented by Yan Zhang, Nilay Noyan, and Roger Lederman
"Accelerating Deliveries: Decomposition-based Approaches for Large-Scale Air Network Optimization", 1:39 - 1:57pm, presented by Tulio Toffolo, Haroldo Santos, Valentina Vaca, Ruilin Ouyang, Wendian Wan, and Na An
Room: Summit - Ballroom 2
Presentations:
"Optimizing Amazon's Network Flow Plan: A Decomposition Approach in Balancing Truck Fill Rates and Labor Utilization", 12:45 - 1:03pm, presented by Hyemin Jeon, Theodoros Pantelidis, and Martin Bagaram
"COPPER: Contract Price Optimization Using Willingness-to-Pay Models for Amazon Freight", 1:03 - 1:21pm, presented by Yebin Tao, Juan Xu, and Roger Lederman
"Air Cargo Revenue Management: A Two-Stage Stochastic Program with Learning-based Recourse", 1:21 - 1:39pm, presented by Yan Zhang, Nilay Noyan, and Roger Lederman
"Accelerating Deliveries: Decomposition-based Approaches for Large-Scale Air Network Optimization", 1:39 - 1:57pm, presented by Tulio Toffolo, Haroldo Santos, Valentina Vaca, Ruilin Ouyang, Wendian Wan, and Na An
Room: Summit - Ballroom 2
Innovative Approaches in Stochastic and Explanatory Modeling
October 20, 12:45 PM - 2:00 PM PDT
Website: Link
Presentations:
"Optimal Solutions with Bounded Inequality", 1:00 - 1:15pm, presented by John Hooker, Ozgun Elci, and Peter Zhang
Room: Summit - 344
Presentations:
"Optimal Solutions with Bounded Inequality", 1:00 - 1:15pm, presented by John Hooker, Ozgun Elci, and Peter Zhang
Room: Summit - 344
A Tutorial on Automated Decomposition Methods for Optimization
October 20, 4:00 PM - 5:15 PM PDT
Website: Link
Presentation:
"Automated Decomposition Software", 4:18 - 4:36pm, presented by Matthew Galati
"Decomposing Problem Data Instead of Problem Formulation: An Application to Assortment Optimization", 4:54 - 5:12pm, presented by Taghi Khaniyev, Kaan Cakiroglu, Ali Ilhan Haliloglu, Elif Sena Isik, and Elif Rana Yoner
Room: Regency - 709
Presentation:
"Automated Decomposition Software", 4:18 - 4:36pm, presented by Matthew Galati
"Decomposing Problem Data Instead of Problem Formulation: An Application to Assortment Optimization", 4:54 - 5:12pm, presented by Taghi Khaniyev, Kaan Cakiroglu, Ali Ilhan Haliloglu, Elif Sena Isik, and Elif Rana Yoner
Room: Regency - 709
Labor Planning at Amazon: Handling Uncertainty
October 20, 4:00 PM - 5:15 PM PDT
Website: Link
Presentations:
"Workforce Composition: Fixed or Flexible Labor?", 4 - 4:18pm, presented by Saba Neyshabouri
"Labor Pooling: Nodal Planning", 4:18 - 4:36pm, presented by Hadi Panahi, Thomas Fillebeen, and Roman Levkin
"Capacity-Aware Flexible Shift Planning for Under-the-Roof Operations at Amazon", 4:36 - 4:54pm, presented by Ramon Auad, Thomas Fillebeen, and Roman Levkin
"Intraday Under-the-Roof Staff Allocation for the Ultra-Fast Delivery at Amazon", 4:54 - 5:12pm, presented by Zeynep Sargut and Ramon Auad
Room: Summit - 427
Presentations:
"Workforce Composition: Fixed or Flexible Labor?", 4 - 4:18pm, presented by Saba Neyshabouri
"Labor Pooling: Nodal Planning", 4:18 - 4:36pm, presented by Hadi Panahi, Thomas Fillebeen, and Roman Levkin
"Capacity-Aware Flexible Shift Planning for Under-the-Roof Operations at Amazon", 4:36 - 4:54pm, presented by Ramon Auad, Thomas Fillebeen, and Roman Levkin
"Intraday Under-the-Roof Staff Allocation for the Ultra-Fast Delivery at Amazon", 4:54 - 5:12pm, presented by Zeynep Sargut and Ramon Auad
Room: Summit - 427
ML and Optimization Driven Fulfillment Operations Management
October 20, 4:00 PM - 5:15 PM PDT
Website: Link
Presentations:
"Warehouse Order Picking and Container Fill Optimization in Amazon Grocery Operations", 4 - 4:18pm, presented by Prem Kumar Viswanathan
"Visual-Assisted Approach to Optimizing Picking Operations in Fulfillment Centers", 4:18 - 4:36pm, presented by Abhisek Mukhopadhyay, SMA Bin Al Islam, Kay Zheng, Elcin Cetinkaya, and Akshay Kurapaty
"Ergonomics Aware Stow Assistance", 4:36 - 4:54pm, presented by Abhisek Mukhopadhyay and Andrew Johnson
"Error-Based Forecast Refinement: Bridging Past Mistakes to Enhance Future Predictions", 4:54 - 5:12pm, presented by Chinmoy Mohapatra, Rohit Malshe, and Abhilasha Katariya
Room: Summit - Ballroom 2
Presentations:
"Warehouse Order Picking and Container Fill Optimization in Amazon Grocery Operations", 4 - 4:18pm, presented by Prem Kumar Viswanathan
"Visual-Assisted Approach to Optimizing Picking Operations in Fulfillment Centers", 4:18 - 4:36pm, presented by Abhisek Mukhopadhyay, SMA Bin Al Islam, Kay Zheng, Elcin Cetinkaya, and Akshay Kurapaty
"Ergonomics Aware Stow Assistance", 4:36 - 4:54pm, presented by Abhisek Mukhopadhyay and Andrew Johnson
"Error-Based Forecast Refinement: Bridging Past Mistakes to Enhance Future Predictions", 4:54 - 5:12pm, presented by Chinmoy Mohapatra, Rohit Malshe, and Abhilasha Katariya
Room: Summit - Ballroom 2
Network Analytics at Amazon Last Mile Logistics
October 20, 4:00 PM - 5:15 PM PDT
Website: Link
Panelists: Natarajan Gautam, Abhilasha Katariya, Dipal Gupta, Chinmoy Mohapatra, and Gokce Kahvecioglu
Room: Summit - 331
Panelists: Natarajan Gautam, Abhilasha Katariya, Dipal Gupta, Chinmoy Mohapatra, and Gokce Kahvecioglu
Room: Summit - 331
Pricing and Yield Management at Amazon
October 20, 4:00 PM - 5:15 PM PDT
Website: Link
Presentations:
"Early Stopping Methods at Pricing Lab", 4 - 4:25pm, presented by Sid Sanghi, Daria Zelenia, James Nordlund, and Mohsen Bayati
"Price Optimization with Neural Thompson Sampling with Dropout", 4:25 - 4:50pm, presented by Hyungjun Lee, Ru Wang, and Laleh Jalali
"Combining ML & OR for Efficient Amazon Locker Utilization", 4:50 - 5:15pm, presented by Samyukta Sethuraman
Room: Summit - 339
Presentations:
"Early Stopping Methods at Pricing Lab", 4 - 4:25pm, presented by Sid Sanghi, Daria Zelenia, James Nordlund, and Mohsen Bayati
"Price Optimization with Neural Thompson Sampling with Dropout", 4:25 - 4:50pm, presented by Hyungjun Lee, Ru Wang, and Laleh Jalali
"Combining ML & OR for Efficient Amazon Locker Utilization", 4:50 - 5:15pm, presented by Samyukta Sethuraman
Room: Summit - 339
Amazon Supply Chain Optimization
October 21, 10:45 AM - 12:00 PM PDT
Website: Link
Presentations:
"Dynamic Package Route Computation at Scale", 10:45 - 11:03am, presented by Yuri Shevchenko, Luciana Buriol, Olivier Durand de Gevigney, Nithin Lingala, and Thomas Helleboid
"Consensus Planning for Capacity Planning: A Market-Based Approach to Coordination at Amazon", 11:03 - 11:21am, presented by Garrett van Ryzin
"Consensus Planning with Primal, Dual, and Proximal Agents", 11:21 - 11:39am, presented by Alvaro Maggiar
"A Primal Recovery Method for Improving Convergence Speed of Distributed Algorithms", 11:57am - 12:15pm, presented by Tetiana Parshakova
Room: Summit - Ballroom 2
Presentations:
"Dynamic Package Route Computation at Scale", 10:45 - 11:03am, presented by Yuri Shevchenko, Luciana Buriol, Olivier Durand de Gevigney, Nithin Lingala, and Thomas Helleboid
"Consensus Planning for Capacity Planning: A Market-Based Approach to Coordination at Amazon", 11:03 - 11:21am, presented by Garrett van Ryzin
"Consensus Planning with Primal, Dual, and Proximal Agents", 11:21 - 11:39am, presented by Alvaro Maggiar
"A Primal Recovery Method for Improving Convergence Speed of Distributed Algorithms", 11:57am - 12:15pm, presented by Tetiana Parshakova
Room: Summit - Ballroom 2
Amazon Flexible Under-the-Roof Capacity Planning
October 21, 8:00 AM - 9:15 AM PDT
Website: Link
Presentations:
"Associate Capacity Prediction", 8 - 8:15am, presented by Esra Sisikoglu and Jeronimo Callejas
"Labor Capacity Management with Surge", 8:15 - 8:24am, presented by Dev Das
"Integrating Associate Preference into Labor Planning: Sentiment Analysis", 8:24 - 8:36am, presented by Rachel Rutkowski, Jeronimo Callejas, Roman Levkin, Thomas Fillebeen, and Martin Savelsbergh
"Integrating Associate Preference into Labor Planning: Structural Modeling", 8:36 - 8:48am, presented by Jeronimo Callejas, Rachel Rutkowski, Roman Levkin, Thomas Fillebeen, and Martin Savelsbergh
"Enhancing Associate Capacity via Shift Reservations", 8:48 - 9am, presented by Evan Wyse
"Amazon Flexible Under-the-Roof Capacity Planning", 9 - 9:12am, presented by Thomas Fillebeen and Roman Levkin
Room: Summit - 427
Presentations:
"Associate Capacity Prediction", 8 - 8:15am, presented by Esra Sisikoglu and Jeronimo Callejas
"Labor Capacity Management with Surge", 8:15 - 8:24am, presented by Dev Das
"Integrating Associate Preference into Labor Planning: Sentiment Analysis", 8:24 - 8:36am, presented by Rachel Rutkowski, Jeronimo Callejas, Roman Levkin, Thomas Fillebeen, and Martin Savelsbergh
"Integrating Associate Preference into Labor Planning: Structural Modeling", 8:36 - 8:48am, presented by Jeronimo Callejas, Rachel Rutkowski, Roman Levkin, Thomas Fillebeen, and Martin Savelsbergh
"Enhancing Associate Capacity via Shift Reservations", 8:48 - 9am, presented by Evan Wyse
"Amazon Flexible Under-the-Roof Capacity Planning", 9 - 9:12am, presented by Thomas Fillebeen and Roman Levkin
Room: Summit - 427
Amazon Last-Mile Resource Optimization
October 21, 8:00 AM - 9:15 AM PDT
Website: Link
Presentations:
"Optimizing Tour Slicing via Shortest Path Formulations", 8 - 8:15am, presented by Michael Wagner, Dipal Gupta, and Rohit Malshe
"Logistics Operations with Combined Deliveries and Pickups", 8:15 - 8:30am, presented by Rohit Malshe, Abhilasha Katariya, Ram Thiruveedhi, Chinmoy Mohapatra
"Newsvendor Model for Capacity Planning", 8:30 - 8:45am, presented by Liron Yedidsion and Rohit Malshe
"Essential Components of Capacity Planning Under Uncertainty", 8:45 - 9am, presented by Ram Thiruveedhi and Abhilasha Katariya
"Cycle Optimization Engine: Maximizing Last Mile Delivery Performance", 9 - 9:15am, presented by Wen Chong, Marc Anderson, and Yimin Liu
Room: Summit - Ballroom 2
Presentations:
"Optimizing Tour Slicing via Shortest Path Formulations", 8 - 8:15am, presented by Michael Wagner, Dipal Gupta, and Rohit Malshe
"Logistics Operations with Combined Deliveries and Pickups", 8:15 - 8:30am, presented by Rohit Malshe, Abhilasha Katariya, Ram Thiruveedhi, Chinmoy Mohapatra
"Newsvendor Model for Capacity Planning", 8:30 - 8:45am, presented by Liron Yedidsion and Rohit Malshe
"Essential Components of Capacity Planning Under Uncertainty", 8:45 - 9am, presented by Ram Thiruveedhi and Abhilasha Katariya
"Cycle Optimization Engine: Maximizing Last Mile Delivery Performance", 9 - 9:15am, presented by Wen Chong, Marc Anderson, and Yimin Liu
Room: Summit - Ballroom 2
Modern Statistics in Social Media Analysis
October 21, 10:45 AM - 12:00 PM PDT
Website: Link
Presentations:
"Inferring Unusual Metrics from Social Media", 10:45 - 11:03am, presented by Alex Gilgur
"Causal Inference Through Privacy Compliant Experimentation that Makes Social Advertising More Relevant", 11:03 - 11:21am, presented by Carlos Avello
"Enhancing Low-Resource LLMs Classification with PEFT and Synthetic Data", 11:21 - 11:39am, presented by Parth Patwa
Room: Regency - 602
Presentations:
"Inferring Unusual Metrics from Social Media", 10:45 - 11:03am, presented by Alex Gilgur
"Causal Inference Through Privacy Compliant Experimentation that Makes Social Advertising More Relevant", 11:03 - 11:21am, presented by Carlos Avello
"Enhancing Low-Resource LLMs Classification with PEFT and Synthetic Data", 11:21 - 11:39am, presented by Parth Patwa
Room: Regency - 602
Practitioners' View on Transportation and Logistics Research
October 21, 10:45 AM - 12:00 PM PDT
Designing Sparse Flexibility Structures: What to do when the Long Chain Fails?
October 21, 12:45 PM - 2:00 PM PDT
Under the Roof Process Design and Optimization
October 21, 12:45 PM - 2:00 PM PDT
Website: Link
Presentations:
"Amazon Logistics Labor Planning Platform and its Eco System", 12:45 - 1:03pm, presented by Herm Li and John Jiang
"Experimental Design and Analysis in Supply Chain Operations", 1:03 - 1:21pm, presented by Brayan Ortiz and Andrew Bruce
"A General Framework for Scheduling Under-the-Roof Processes in Amazon", 1:21 - 1:39pm, presented by Nikos Lappas
"From Science to Product: A Case Study on Labor Planning Platform of Amazon Logistics", 1:39 - 1:57pm, presented by Herm Li and Zhengqian Jiang
Room: Summit - Ballroom 2
Presentations:
"Amazon Logistics Labor Planning Platform and its Eco System", 12:45 - 1:03pm, presented by Herm Li and John Jiang
"Experimental Design and Analysis in Supply Chain Operations", 1:03 - 1:21pm, presented by Brayan Ortiz and Andrew Bruce
"A General Framework for Scheduling Under-the-Roof Processes in Amazon", 1:21 - 1:39pm, presented by Nikos Lappas
"From Science to Product: A Case Study on Labor Planning Platform of Amazon Logistics", 1:39 - 1:57pm, presented by Herm Li and Zhengqian Jiang
Room: Summit - Ballroom 2
AI at Amazon Last-Mile
October 21, 2:15 PM - 3:30 PM PDT
Website: Link
Presentations:
"Revolutionizing Last Mile Delivery Through Channel Topology Optimization", 2:15 - 2:33pm, presented by Jin Ye and Gokce Kahvecioglu
"Large Scale Events", 2:33 - 2:51pm, presented by Selin Tosun, Jin Ye, and Leah Riley
"Optimizing Last MIle Delivery with Machine Learning: Amazon's Infrastructure and Scaling Strategies", 2:51 - 3:09pm, presented by Xiaodong Lan, Eliot Hijano, Ram Thiruveedhi, Abhilasha Katariya, and Rohit Malshe
"Cold-Start Forecasting for Special Events", 3:09 - 3:27pm, presented by Abhilasha Katariya, Ram Thiruveedhi, and Arkajyoti Misra
Room: Summit - Ballroom 2
Presentations:
"Revolutionizing Last Mile Delivery Through Channel Topology Optimization", 2:15 - 2:33pm, presented by Jin Ye and Gokce Kahvecioglu
"Large Scale Events", 2:33 - 2:51pm, presented by Selin Tosun, Jin Ye, and Leah Riley
"Optimizing Last MIle Delivery with Machine Learning: Amazon's Infrastructure and Scaling Strategies", 2:51 - 3:09pm, presented by Xiaodong Lan, Eliot Hijano, Ram Thiruveedhi, Abhilasha Katariya, and Rohit Malshe
"Cold-Start Forecasting for Special Events", 3:09 - 3:27pm, presented by Abhilasha Katariya, Ram Thiruveedhi, and Arkajyoti Misra
Room: Summit - Ballroom 2
Advancement in High-dimensional Data Analytics
October 21, 4:00 PM - 5:15 PM PDT
Website: Link
Presentation: "High Dimensional Packages' Delivery Times Prediction under Aggregated Quantile Constraints", presented by Yifei Yuan, Mederic Motte, and Philip Kaminsky
Room: Summit - 322
Presentation: "High Dimensional Packages' Delivery Times Prediction under Aggregated Quantile Constraints", presented by Yifei Yuan, Mederic Motte, and Philip Kaminsky
Room: Summit - 322
Amazon Geospatial Planning
October 21, 4:00 PM - 5:15 PM PDT
Website: Link
Presentations:
"Optimizing Van Loading and Dispatching Operations in Amazon's Last Mile Delivery Stations", 4 - 4:18pm, presented by Yaniv Mordecai, Liron Yedidsion, and Rohit Malshe
"Geospatial Optimization at Amazon", 4:18 - 4:36pm, presented by Dipal Gupta
"Last Mile Drivers Capacity Scaling", 4:36 - 4:54pm, presented by Mahdieh Allahviranloo and Jin Ye
"Time Series Ensemble Forecasting Models", 4:54 - 5:12pm, presented by Abhilasha Katariya and Eliot Hijano
Room: Summit - Ballroom 2
Presentations:
"Optimizing Van Loading and Dispatching Operations in Amazon's Last Mile Delivery Stations", 4 - 4:18pm, presented by Yaniv Mordecai, Liron Yedidsion, and Rohit Malshe
"Geospatial Optimization at Amazon", 4:18 - 4:36pm, presented by Dipal Gupta
"Last Mile Drivers Capacity Scaling", 4:36 - 4:54pm, presented by Mahdieh Allahviranloo and Jin Ye
"Time Series Ensemble Forecasting Models", 4:54 - 5:12pm, presented by Abhilasha Katariya and Eliot Hijano
Room: Summit - Ballroom 2
Measuring the Efficacy of Amazon’s Recommendation Systems
October 21, 4:00 PM - 5:15 PM PDT
Website: Link
Speakers: Ozalp Ozer, Serdar Simsek, Xiaoxi Zhao, Ethan Dee, and Vivian Yu
Room: Summit - Signature Room
Abstract:
Amazon’s Fulfillment By Amazon (FBA) program provides assistance to Selling Partners (“sellers,” for short) in the form of information sharing, recommendations guiding seller actions (e.g., restock quantity recommendations, excess inventory recommendations), and delegated actions (e.g., automated removals of aged inventory). Amazon’s vision is to help sellers make better decisions and achieve better business outcomes.
In this tutorial, we consider the sophisticated optimization models Amazon employs to generate recommendations. For example, if a seller has excess inventory, Amazon recommends actions to increase their sell-through rate, such as creating a sale or Sponsored Product ad. We demonstrate how we measure the efficacy of these recommendation systems on seller-product outcomes (e.g., revenue, units shipped, and customer clicks on product listings, or “glance views”). Measuring such outcomes is a causal inference problem because we only observe each seller-product’s “factual” and not their “counterfactual” outcome. We employ causal machine learning methodologies such as double machine learning, causal forest, and doubly-robust forest to separate selection bias from a comparison of “treatment” and “control” sellers. For example, we find that aligning with the restock and excess inventory recommendations, on average, improves several seller-salient outcomes. We also present methods for measuring heterogeneity in the efficacy of these recommendations across seller and product segments, and estimate personalized benefits for each seller-product. Finally, through A/B testing, we find that sharing quantified efficacy information with sellers increases their adoption of Amazon recommendations. Sellers are responding to this messaging, and the duty to them is to rigorously identify causal estimates.
Speakers: Ozalp Ozer, Serdar Simsek, Xiaoxi Zhao, Ethan Dee, and Vivian Yu
Room: Summit - Signature Room
Abstract:
Amazon’s Fulfillment By Amazon (FBA) program provides assistance to Selling Partners (“sellers,” for short) in the form of information sharing, recommendations guiding seller actions (e.g., restock quantity recommendations, excess inventory recommendations), and delegated actions (e.g., automated removals of aged inventory). Amazon’s vision is to help sellers make better decisions and achieve better business outcomes.
In this tutorial, we consider the sophisticated optimization models Amazon employs to generate recommendations. For example, if a seller has excess inventory, Amazon recommends actions to increase their sell-through rate, such as creating a sale or Sponsored Product ad. We demonstrate how we measure the efficacy of these recommendation systems on seller-product outcomes (e.g., revenue, units shipped, and customer clicks on product listings, or “glance views”). Measuring such outcomes is a causal inference problem because we only observe each seller-product’s “factual” and not their “counterfactual” outcome. We employ causal machine learning methodologies such as double machine learning, causal forest, and doubly-robust forest to separate selection bias from a comparison of “treatment” and “control” sellers. For example, we find that aligning with the restock and excess inventory recommendations, on average, improves several seller-salient outcomes. We also present methods for measuring heterogeneity in the efficacy of these recommendations across seller and product segments, and estimate personalized benefits for each seller-product. Finally, through A/B testing, we find that sharing quantified efficacy information with sellers increases their adoption of Amazon recommendations. Sellers are responding to this messaging, and the duty to them is to rigorously identify causal estimates.
Pricing and Revenue Management Problems in Supply Chain Networks
October 21, 4:00 PM - 5:15 PM PDT
Website: Link
Presentations:
"Bandit-Based Assortment Strategies with Application to Ultra-Fast Delivery", 4 - 4:18pm, presented by Hunyong Cho, Jonathan Jonker, Lina Al-Kanj, Boran Hu, Greg Herman, Josiah Davis, Zachary Hervieux-Moore, Eric Laber, and Siyeon Kim
"Dynamic Pricing via Reinforcement Learning in a Spot Marketplace", 4:18 - 4:36pm, presented by Jacob Tutmaher, Huiwen Jia, Priyanka Shende, Mohsen Moarefdoost, and Philip Kaminsky
"Offer Selections Optimizer in Transportation Capacity", 4:36 - 4:52pm, presented by Mohsen Moarefdoost, Huiwen Jia, Philip Kaminsky, and Yaser Ghaedsharaf
"Spot Market Pricing on Amazon Freight", 4:54 - 5:12pm, presented by Goutam Kumar, Adam Elmachtoub, and Roger Lederman
Room: Summit - 343
Presentations:
"Bandit-Based Assortment Strategies with Application to Ultra-Fast Delivery", 4 - 4:18pm, presented by Hunyong Cho, Jonathan Jonker, Lina Al-Kanj, Boran Hu, Greg Herman, Josiah Davis, Zachary Hervieux-Moore, Eric Laber, and Siyeon Kim
"Dynamic Pricing via Reinforcement Learning in a Spot Marketplace", 4:18 - 4:36pm, presented by Jacob Tutmaher, Huiwen Jia, Priyanka Shende, Mohsen Moarefdoost, and Philip Kaminsky
"Offer Selections Optimizer in Transportation Capacity", 4:36 - 4:52pm, presented by Mohsen Moarefdoost, Huiwen Jia, Philip Kaminsky, and Yaser Ghaedsharaf
"Spot Market Pricing on Amazon Freight", 4:54 - 5:12pm, presented by Goutam Kumar, Adam Elmachtoub, and Roger Lederman
Room: Summit - 343
Amazon Warehouse Resource Optimization
October 22, 8:00 AM - 9:15 AM PDT
Website: Link
Presentation:
"Labor Staffing in Amazon Sortable Fulfillment Centers" presented by Yingqiu Zhang
"Real-TIME Staffing Plan Adjustment in Fulfillment Centers" presented by Xinyu Fei and Andrew Johnson
"Same Day Outbound Resource Planning" presented by Aaron Herman
"Outbound Flow Management Systems" presented by Maximilian Zellner
"Modeling Flow in Fulfillment Centers" presented by Maximilian Zellner and Andrew Johnson
Room: Summit - Ballroom 2
Presentation:
"Labor Staffing in Amazon Sortable Fulfillment Centers" presented by Yingqiu Zhang
"Real-TIME Staffing Plan Adjustment in Fulfillment Centers" presented by Xinyu Fei and Andrew Johnson
"Same Day Outbound Resource Planning" presented by Aaron Herman
"Outbound Flow Management Systems" presented by Maximilian Zellner
"Modeling Flow in Fulfillment Centers" presented by Maximilian Zellner and Andrew Johnson
Room: Summit - Ballroom 2
Capacity Planning for Amazon Customer Service Contact Center
October 22, 8:00 AM - 9:15 AM PDT
Website: Link
Presentations:
"Prediction Interval in Demand Forecasting", 8 - 8:15am, presented by Stephanie Mao
"Tesseract - Forecasting Automation Solutions", 8:15 - 8:30am, presented by Chen Liang and Yongfeng Hui
"A Doubly Stochastic Poisson Arrival Model for Amazon Contact Center: Performance Evaluation and Staffing Implications", 8:30 - 8:45am, presented by Kevin Melendez and Xin Liu
"Short-Term Planning in Amazon Customer Service", 8:45 - 9am, presented by Adolfo Rocco and Kevin Melendez
"Estimating Interval-Based Tail Probability of Delay in Real Time", 9 - 9:15am, presented by Ling Zhang and Yunan Liu
Room: Summit - 435
Presentations:
"Prediction Interval in Demand Forecasting", 8 - 8:15am, presented by Stephanie Mao
"Tesseract - Forecasting Automation Solutions", 8:15 - 8:30am, presented by Chen Liang and Yongfeng Hui
"A Doubly Stochastic Poisson Arrival Model for Amazon Contact Center: Performance Evaluation and Staffing Implications", 8:30 - 8:45am, presented by Kevin Melendez and Xin Liu
"Short-Term Planning in Amazon Customer Service", 8:45 - 9am, presented by Adolfo Rocco and Kevin Melendez
"Estimating Interval-Based Tail Probability of Delay in Real Time", 9 - 9:15am, presented by Ling Zhang and Yunan Liu
Room: Summit - 435
Supply Chain Planning and Execution
October 22, 10:45 AM - 12:00 PM PDT
Website: Link
Presentations:
"Validating Networking Planning via Customer Centric Baselines", 10:45 - 11am, presented by Qi Chen
"Middle Mile Capacity Planning", 11 - 11:15am, presented by Ioannis Spantidakis
"Causal Inference in Transportation Efficiency Estimation", 11:15 - 11:30am, presented by Zhikun Gao
"Inventory Feasible and Capacity Compliant Facility Assignment Model" presented by Ayten Turkcan Upasani, Arun Jotshi, Yufei Wang, Kaushik Krishnan
"Solving Large Scale Assignment Problems with Decomposition", 11:45am - 12pm, presented by Ozlem Bilginer
Room: Summit - Ballroom 2
Presentations:
"Validating Networking Planning via Customer Centric Baselines", 10:45 - 11am, presented by Qi Chen
"Middle Mile Capacity Planning", 11 - 11:15am, presented by Ioannis Spantidakis
"Causal Inference in Transportation Efficiency Estimation", 11:15 - 11:30am, presented by Zhikun Gao
"Inventory Feasible and Capacity Compliant Facility Assignment Model" presented by Ayten Turkcan Upasani, Arun Jotshi, Yufei Wang, Kaushik Krishnan
"Solving Large Scale Assignment Problems with Decomposition", 11:45am - 12pm, presented by Ozlem Bilginer
Room: Summit - Ballroom 2
Data-Driven Supply Chain Optimization: Demand Forecasting and Deep Reinforcement
October 22, 1:25 PM - 2:00 PM PDT
Website: Link
Presenter: Sohrab Andaz
Abstract:
Inventory control is a complex real-world problem that involves handling challenges like seasonal demand, time-varying costs, complex inventory arrival dynamics, and various real-world constraints. In modern supply chain management, effective inventory control requires more than traditional optimization methods. Over the past few years, the Amazon SCOT Demand Forecasting and RL groups have pioneered deep learning models that distill Amazon-scale data into deployable time-series forecasting models and control policies designed to handle these complexities.
In this talk, we will explore both approaches, demonstrating:
Presenter: Sohrab Andaz
Abstract:
Inventory control is a complex real-world problem that involves handling challenges like seasonal demand, time-varying costs, complex inventory arrival dynamics, and various real-world constraints. In modern supply chain management, effective inventory control requires more than traditional optimization methods. Over the past few years, the Amazon SCOT Demand Forecasting and RL groups have pioneered deep learning models that distill Amazon-scale data into deployable time-series forecasting models and control policies designed to handle these complexities.
In this talk, we will explore both approaches, demonstrating:
- How deep learning techniques naturally handle supply chain forecasting challenges, including seasonality, cold starts, diverse product categories, and forecast volatility.
- How deep reinforcement learning can move inventory control beyond the predict-then-optimize framework, allowing practitioners to directly optimize business objectives using historical data.
Inventory Placement and Inbound Network Design
October 22, 2:15 PM - 3:30 PM PDT
Website: Link
Presentations:
"Probabilistic Approach to Large-Scale Inventory Placement", 2:15 - 2:33pm, presented by Cristiana Lara, Arash Haddadan, David Mildebrath, and R Ravi
"Evaluating Regional Decentralization of Inventory in Delivery Speed-Sensitive Retail Networks: A Comparative Study between Newsvendor Approximation and Simulation", 2:33 - 2:51pm, presented by Katja Meuche, Yuan Li, Amitabh Sinha, and Benoit Montreuil
"A Unified Bulk Storage and Distribution Network Design", 2:51 - 3:09pm, presented by Shanshan Zhang
"Ideal Inventory Placement with Demand Spillover", 3:09 - 3:27pm, presented by Zihao Li, Zhongxiang Wang, and Tolga Cezik
Room: Summit - Ballroom 2
Presentations:
"Probabilistic Approach to Large-Scale Inventory Placement", 2:15 - 2:33pm, presented by Cristiana Lara, Arash Haddadan, David Mildebrath, and R Ravi
"Evaluating Regional Decentralization of Inventory in Delivery Speed-Sensitive Retail Networks: A Comparative Study between Newsvendor Approximation and Simulation", 2:33 - 2:51pm, presented by Katja Meuche, Yuan Li, Amitabh Sinha, and Benoit Montreuil
"A Unified Bulk Storage and Distribution Network Design", 2:51 - 3:09pm, presented by Shanshan Zhang
"Ideal Inventory Placement with Demand Spillover", 3:09 - 3:27pm, presented by Zihao Li, Zhongxiang Wang, and Tolga Cezik
Room: Summit - Ballroom 2
Large Scale Supply Chain Network Design
October 22, 2:15 PM - 3:30 PM PDT
Website: Link
Presentations:
"Solving the Continuous TIME Service Network Design Problem (CTSND) by Column Generation", 2:15 - 2:33pm, presented by Marcus Poggi, Arash Haddadan, Daniel Ulch, Jochen Koenemann, and Madison Van Dyk
"Designing Delivery-Anchors for Amazon's Middle Mile Network", 2:33 - 2:51pm, presented by Arash Haddadan, Jochen Koenemann, Leian Chen, Mityansh Seth
"Designing Optimzation Software for Use-Case Scalability", 2:51 - 3:09pm, presented by Semih Atakan
"Region-Based Network Design", 3:09 - 3:27pm, presented by Baris Burnak, Semih Atakan, and Jochen Koenemann
Room: Summit - 421
Presentations:
"Solving the Continuous TIME Service Network Design Problem (CTSND) by Column Generation", 2:15 - 2:33pm, presented by Marcus Poggi, Arash Haddadan, Daniel Ulch, Jochen Koenemann, and Madison Van Dyk
"Designing Delivery-Anchors for Amazon's Middle Mile Network", 2:33 - 2:51pm, presented by Arash Haddadan, Jochen Koenemann, Leian Chen, Mityansh Seth
"Designing Optimzation Software for Use-Case Scalability", 2:51 - 3:09pm, presented by Semih Atakan
"Region-Based Network Design", 3:09 - 3:27pm, presented by Baris Burnak, Semih Atakan, and Jochen Koenemann
Room: Summit - 421
Constraints and Coordination for Deep Reinforcement Learning Agents
October 22, 2:55 PM - 3:30 PM PDT
Website: Link
Presenter: Carson Eisenach
Abstract:
We explore new techniques for constrained reinforcement learning (RL) in the real-world, specifically as applied to inventory management. The classic approach is to use model predictive control to enforce constraint adherence. With Deep RL policies, this becomes complicated as they consume high-dimensional features to make decisions, and accurately forward simulating this joint distribution is extremely challenging, if not impossible. This session provides an overview of a new approach — “Neural Coordinator” — which directly forecasts dual costs under which a policy will adhere to the desired constraints. We will demonstrate its effectiveness in the inventory control setting and cover how to back-test policies in the presence of constraints.
Presenter: Carson Eisenach
Abstract:
We explore new techniques for constrained reinforcement learning (RL) in the real-world, specifically as applied to inventory management. The classic approach is to use model predictive control to enforce constraint adherence. With Deep RL policies, this becomes complicated as they consume high-dimensional features to make decisions, and accurately forward simulating this joint distribution is extremely challenging, if not impossible. This session provides an overview of a new approach — “Neural Coordinator” — which directly forecasts dual costs under which a policy will adhere to the desired constraints. We will demonstrate its effectiveness in the inventory control setting and cover how to back-test policies in the presence of constraints.
Logistics and Network Optimization Challenges
October 22, 4:00 PM - 5:15 PM PDT
Website: Link
Presentation: "Execution Time Estimation for Amazon Last Mile Route Planning", 4 - 4:12pm, presented by Bharath Veluri and Duo Zheng
Room: Summit - 344
Presentation: "Execution Time Estimation for Amazon Last Mile Route Planning", 4 - 4:12pm, presented by Bharath Veluri and Duo Zheng
Room: Summit - 344
AI and Optimization Techniques in Forecasting and Security
October 22, 4:00 PM - 5:15 PM PDT
Website: Link
Presentation:
"Hierarchical Forecasting for Amazon EC2 Demand", 4:18 - 4:36pm, presented by Alex (Yiming) Wang and Ebrahim Nasrabadi
Room: Summit - 328
Presentation:
"Hierarchical Forecasting for Amazon EC2 Demand", 4:18 - 4:36pm, presented by Alex (Yiming) Wang and Ebrahim Nasrabadi
Room: Summit - 328
