Retail AI Data Engineering: How Naresh Erukulla Drives Macy's Forecasting

Lead Data Engineer Naresh Erukulla details how modern platforms at Macy's use AI orchestration to improve retail forecasting. By building self-healing systems and integrating LLM agents, the company achieved USD 100 million in savings. This approach ensures data reliability, helping retailers manage complex inventory and supply chain decisions through high-quality, auditable insights and automated planning.

As artificial intelligence moves from experimental tools to core operational infrastructure, the real bottleneck is no longer model innovation but the systems required to support it. For large enterprises, especially retailers managing millions of products and billions of signals, the question has shifted from “What can AI do?” to “How do we engineer an environment where AI can operate reliably, continuously, and at scale?”

Naresh Erukulla, a Lead Data Engineer at Macy’s, guiding next generation forecasting and AI orchestration inside Macy’s, has spent his career answering that question. His work reflects a belief that the future of decision making rests on data systems that heal themselves, scale predictably, and integrate seamlessly with the cognitive layer of AI driven agents. It is a perspective shaped not only by industry execution, but also by his thought leadership as a DZone author of Self Healing Data Pipelines: The Next Big Thing in Data Engineering? “AI does not fail because models are weak. It fails when the systems around the model cannot support the weight of real world complexity,” he says. “The real frontier in enterprise AI is reliability.”

This conviction has shaped the engineering strategy behind Macy’s preseason forecasting engine, a program now central to the company’s inventory, merchandising, and supply chain decisions.

The Rise of AI Enabled Planning in Enterprise Retail

Retail operates in the middle of macro forces that are increasingly unpredictable. Climate patterns shift seasonal demand. Global logistics disruptions reshape distribution timelines. Consumer sentiment can change within days. In this environment, traditional forecasting tools that rest on historical data alone struggle to anticipate volatility. The most successful retailers are therefore moving toward intelligent frameworks that combine structured metrics with unstructured signals, such as merchant notes, customer sentiment, and local market insights.

The transformation is structural rather than cosmetic. AI driven forecasting does not simply improve accuracy. It changes how organizations communicate, plan, and negotiate inventory decisions across thousands of stores and digital channels. Macy’s recognized this shift early and invested in a forecasting ecosystem that could think ahead rather than react backward. Naresh became one of the principal architects shaping that infrastructure.

“Our job is to give the business a view of the future that is both quantitative and contextual,” Naresh explains. “AI can forecast the curve, but data engineering determines whether the insights are trustworthy enough to act on.”

Building the System That Allows AI to Think

The platform Naresh helped design predicts demand months ahead, synthesizing vast seasonal, operational, and behavioral datasets. It sits on top of a lakehouse architecture he engineered to ingest, transform, and validate billions of data points flowing from POS systems, third party signals, inventory feeds, weather patterns, and merchandising metadata. This foundation ensures that both ML models and LLM agents receive inputs that are fresh, consistent, and auditable.

The forecasting engine layered on top of this backbone integrates multiple families of models, including gradient boosting, deep neural networks, LSTMs, and transformer architectures. What makes the platform distinct is the AI agent orchestration environment that Naresh shaped: a system where LLMs interpret natural language questions, run scenario simulations, reconcile forecasting hierarchies, and deliver explanations that planners can use in real time.

“People do not want a black box. They want a conversation,” Naresh says. “The real innovation is when AI can respond contextually, explain its reasoning, and adapt to new information instantly.”

The technical lesson is clear. AI becomes useful only when paired with a robust orchestration layer capable of managing data quality, pipeline resilience, and continuous refresh cycles. Naresh’s engineering leadership turned these principles into repeatable frameworks now used across Macy’s broader data ecosystem.

From Architecture to Business Impact

AI enabled forecasting has emerged as one of the largest financial levers in modern retail. Macy’s has publicly tied its inventory improvements to advanced forecasting and automation, reporting a fifteen percent gain in inventory turnover and outlining more than one hundred million dollars in expected supply chain savings tied to data driven decision systems.

The forecasting platform Naresh helped design plays a central role in these gains. With more accurate preseason projections, the company reduces excess inventory, stabilizes working capital, and minimizes markdown risk. Faster scenario analysis enables planners and merchants to adjust buying strategies within minutes rather than days. Highly automated fulfillment centers such as China Grove rely on these predictions to orchestrate nationwide replenishment.

“Speed matters, but confidence matters more,” Naresh reflects. “When the architecture protects the integrity of the forecast, the business becomes far more agile.”

Operating Principles That Scale

Beyond raw engineering, Naresh is an author of a scholarly publication title Efficient Orchestration of AI Workloads: Data Engineering Solutions for Distributed Cloud Computing. He emphasizes an operating discipline that makes innovation repeatable. He develops systems built for self healing, a philosophy reflected in his writing on next generation data pipelines. This approach reduces human intervention, lowers maintenance overhead, and shields downstream teams from disruptions. By standardizing pipeline design, quality validations, and lineage frameworks, he ensures that new AI tools can be adopted without destabilizing existing workflows.

His leadership has shaped Macy’s modernization efforts, earning recognition internally for his ability to align data engineering with long term strategic priorities. For Naresh, success is measured not only in technical metrics but in organizational clarity.

“A great data system is invisible,” he explains. “It works so reliably that people stop worrying about the plumbing and focus entirely on decisions.”

Looking Ahead: The Future of Autonomous Planning

As AI becomes embedded in enterprise operations, retailers will continue moving toward autonomous planning systems that partner with human decision makers. In the years ahead, Naresh,a judge for the Globee Awards for Leadership, anticipates that LLM agents will evaluate risks, propose buys, generate reconciled recommendations, and negotiate tradeoffs before planners ever enter a meeting. The engineering foundations he has built position Macy’s for that next era.

“AI will not replace strategic judgment,” he says. “It will enhance it by giving people the clearest possible picture of what the future might look like.”

In a landscape where information is abundant but usable intelligence is rare, Naresh represents the emerging generation of data leaders who convert complexity into lasting enterprise advantage, proving that the future of retail belongs to organizations where architecture and intelligence evolve together.