Baljeet Arora

Demand Science Optimization, Amazon Devices and Services

• Led technical strategy and engineering execution to build distributed systems for Amazon Devices’ demand forecasting, pricing, and inventory allocation. Drove multiple SVP/VP-level goals to deliver over $200 million in operating income across global retail channels through improved decision-making. Led cross-functional teams of over 100 to deliver GenAI/LLM applications, large-scale OR/ML systems, and global ML Ops infrastructure.
• Architected and built Amazon Devices’ central modeling infrastructure—a shared platform with parallelized execution and standardized interfaces. Reduced time-to-market for science models by 45%.
• Launched an elasticity-based pricing experimentation platform to run price tests directly in production, enabling near real-time pricing decisions based on demand and promotional constraints.
• Built and deployed automated device pricing and promotion algorithms. Upgraded planning systems by launching a unified demand signal, and integrating real-time scenario planning and simulation tools.
• Led SVP/VP-level cross-organizational initiative to improve engineering and operational excellence, including GenAI tools that improved developer productivity by 10%.
• Established and scaled an engineering center in Mexico City; hired and ramped up a 30-person SDE team, reducing employee costs by 10% while maintaining global development standards.

Fulfillment Optimization, Supply Chain Optimization Technologies (SCOT)

• Led cross-functional engineering and science teams (of 50+) to optimize network routing decisions for Amazon’s fulfillment planning systems, handling billions of packages and customer orders globally. Developed a scalable solution to simulate order assignment behavior and automate millions of network configurations, saving over $100M in transportation costs. This project was featured in Amazon’s letter to the shareholders.
• Led the development of optimized network design systems that improved delivery efficiency by 18% for millions of packages annually. Drove a strategic initiative that optimized network routing by building efficient network designs globally.
• Designed and launched a network connectivity optimizer to balance delivery speed, cost, and sustainability, using large-scale optimization, problem decomposition, network flow structures, and column generation techniques.
• Built science models with operations research, predictive analytics, and graph databases to enhance the outbound network. Deployed these models into the AWS cloud-based infrastructure. Work was presented at INFORMS, and is patent pending.

Supply Chain Optimization Technology (SCOT)

• Built and scaled analytical and simulation frameworks to drive strategic decision-making across the supply chain. Developed statistical and economic models that directly influenced network planning, inventory optimization, and transportation efficiency, delivering significant cost savings and operational improvements.
• Developed an end-to-end supply chain simulation system referenced in the book The Everything Store, enabling scenario planning and data-driven decisions at scale.
• Designed inventory allocation and order planning models that reduced shipping costs and improved delivery speed, contributing to over $200 million in savings.
• Developed statistical/OR models to determine which new Amazon fulfillment centers we should open to make the business profitable. This led to the savings of over $50 million per year in cost savings by right-sizing the fulfillment network.