Online grocery orders surged during the pandemic, testing Walmart’s supply chain in ways traditional forecasting models couldn’t handle. Demand patterns shifted unpredictably, departments operated based on conflicting projections, and the company needed a system that could adapt in real-time. Shashank Kadetotad, then Associate Director of Advanced Analytics for Supply Chain, rebuilt the forecasting infrastructure from the ground up, work that would become a blueprint for his later enterprise AI strategies at Mars.
Kadetotad’s 16-year career across Amazon, Walmart, and Mars traces an unusual trajectory. He launched Amazon’s first fulfillment centers in India and Poland, where he learned to build systems flexible enough for markets where address standardization didn’t yet exist. He spent years at Walmart turning predictive models into billion-dollar operational improvements. Today, he directs generative AI strategy at Mars, where algorithms co-create decisions alongside human judgment. Each chapter taught the same lesson: technical solutions fail when they ignore organizational reality.
Building Systems That Speak One Language
Walmart’s eCommerce network relied on separate forecasting models for Finance, Operations, and Supply Chain before Kadetotad arrived. These siloed approaches yielded conflicting projections, creating inefficiencies in inventory allocation and labor planning. Finance sought predictability, the supply chain was optimized for efficiency, operations focused on execution, and marketing sought agility. Reconciling those competing priorities required more than mathematics.
Kadetotad introduced “One Forecast,” a centralized system that integrated data from all functions into a single, enterprise-wide model. The challenge wasn’t data or algorithms, it was trust. “The challenge wasn’t data or models—it was trust,” he recalls. “We had to build a common language of truth that all functions could rally around.” The new system incorporated machine learning enhancements and continuous model tuning, resulting in over 25% year-over-year improvement in forecast accuracy. Tracking granular metrics, such as fulfillment center cube consumption, backlog resolution, and delivered units, provided the team with real-time visibility into network performance.
The pilot phase delivered $8 million in measurable value within six months, justifying expansion across the entire company. That principle, organizational change scales through inclusion rather than imposition, would guide Kadetotad’s later work at Mars, where he focuses as much on human alignment as technical excellence. Technology accelerates intelligence, but culture sustains it.
From Rearview Mirror to Windshield
Accurate forecasting solved only part of the problem. Kadetotad led the development of a simulation tool that modeled various operational scenarios, from vendor compliance changes to regional demand spikes. Teams could make proactive adjustments in labor, transportation, and inventory, minimizing disruptions before they materialized.
The On-Time In-Full (OTIF) metric, a critical measure of delivery performance, improved by 20% year-over-year, maintaining levels above 95% even during peak seasons. “We stopped treating data as a rearview mirror,” Kadetotad explains. “The goal was to make it a windshield—something that helps you see where you’re headed.” That shift from reactive reporting to forward-looking guidance set a new standard for decision-making within the organization.
Walmart represented the middle stage of an evolution Kadetotad has tracked from inside the room where decisions get made. Early Amazon systems informed you about what happened after the fact—descriptive analytics that measured operational success. Walmart marked a shift toward predictive models that optimized supply chains and demand forecasts before problems materialized. Mars represents the next frontier: generative AI that not only guides decisions but also co-creates them, simulating scenarios and generating insights previously unimaginable.
Three Cultures, One Philosophy
Each Fortune 500 company shaped a different dimension of Kadetotad’s leadership approach. Amazon taught inventive urgency and building mechanisms that scale experimentation. Walmart demonstrated how the smallest optimization can create billion-dollar impacts when applied at a massive scale. Mars, a family-owned and generation-minded company, emphasized that decisions should serve purposes that extend beyond just quarterly earnings.
Those environments produced a hybrid philosophy: driving bold technological change while keeping people and long-term value at the center. “It’s about scaling intelligence, not just algorithms,” he notes. His educational background positioned him for this translator role, electrical engineering taught him how systems think, while his MBA from the University of Michigan’s Ross School of Business taught him how organizations think. “It’s not enough to understand models; you have to understand markets, ethics, and human behavior,” he argues.
Speaking at conferences from Generative AI Week to DataIQ summits, Kadetotad sees patterns in how organizations stumble. The most common mistake: treating AI as isolated pilots that never reach production because they lack enterprise architecture and clear value narratives. Another pitfall: over-indexing on technology while underinvesting in change management. Employees must trust AI-driven insights enough to act on them, requiring transparency, education, and responsible design.
Companies that will dominate the coming decade treat AI not as a tool but as a transformation of how decisions get made, where machine intelligence becomes invisible because it’s woven seamlessly into how business thinks, acts, and learns. Kadetotad’s systems at Walmart laid the foundation for his current work at Mars, where he directs the Global Senior Director of Enterprise Data Science and AI Delivery. He has established an AI Lab that has filed five provisional patents, built MarsGPT (which has generated $7 million in value), and implemented a Responsible AI framework that has become a company-wide program.
His trajectory from fulfillment warehouses to generative AI patents suggests that the most effective technical leaders often bring deep operational understanding to their strategies. They know what breaks, what people resist, what actually needs solving—knowledge earned in markets where infrastructure hadn’t caught up to ambition, in cross-functional battles where competing definitions of success had to be reconciled, in cultures that value both speed and sustainability. The DataIQ award winner continues to speak at industry events and direct Mars’ AI transformation, demonstrating that careers built on diverse operational experiences can culminate in technical thought leadership.
