Across the retail sector, the competitive frontier is shifting from who captures data to who can transform that data into decisions at enterprise scale. Forecasting cycles that were once annual rituals built on static spreadsheets now operate in a fluid environment where profit depends on how well a company interprets billions of signals in real time. As retailers accumulate oceans of data, the real differentiator lies in the infrastructure behind the scenes: the pipelines, lakehouses, AI agents, and engineering systems that allow intelligence to function reliably across thousands of internal decisions.
Naresh Erukulla, a Lead Data Engineer at Macy’s, and an IEEE Panel Reviewer who has shaped some of the most advanced forecasting and AI orchestration systems, works at this intersection of scale and intelligence. He approaches engineering with a principle he has repeated throughout his career: simplify complexity, preserve clarity, and build systems that raise the decision quality of an entire organization. “AI succeeds when infrastructure removes friction,” he says. “The real question is how quickly and confidently a business can act once its data is organized, interpretable, and ready for intelligence.”
This mindset has guided his work at Macy’s, where he helped design one of the company’s most ambitious transformations in long range forecasting and inventory planning.
Retail today is defined by volatility. Weather anomalies, global supply chain shifts, rapid trend cycles, and unpredictable consumer sentiment have made traditional forecasting insufficient. The United States retail landscape increasingly follows the operational standard set by technology driven leaders like Walmart and Amazon, both of which have invested heavily in AI based forecasting, automated allocation systems, and intelligence driven supply networks.
In this climate, forecasting is no longer an analytical exercise. It is a strategic business function that determines capital deployment, inventory availability, and margin protection. Retailers who modernize the intelligence layer behind their planning engines are discovering that the value does not simply come from more accurate predictions. It emerges from the speed at which organizations can simulate scenarios, debate options, and move inventory with confidence.
“Demand signals change constantly,” Naresh explains. “The advantage comes from understanding those shifts early and acting before risk turns into cost.”
This belief shaped the architecture of Macy’s Preseason Projection Platform, a unified forecasting and AI engine that redefines the company’s planning capabilities.
The platform Naresh designed predicts demand for every store SKU combination up to a year in advance and supports strategic decisions around buying, allocation, and open to buy planning. It incorporates advanced machine learning, deep learning, and a novel layer of LLM powered retail agents capable of interpreting unstructured data such as planner notes, emerging trend reports, sentiment patterns, and local store insights.
The data backbone behind this system includes real time ingestion frameworks, a medallion modeled lakehouse, a high performance feature store, and automated quality and lineage systems. These foundations deliver the stability needed for the forecasting engine, which uses gradient boosted models, deep neural networks, LSTMs, and transformer based sequence architectures to analyze complex seasonal patterns.
The differentiator is the AI agent layer. These agents interpret natural language questions from planners, run multi scenario simulations, orchestrate model calls, and generate recommendations that normally require teams of analysts. This bridge between qualitative expertise and quantitative precision is a turning point in enterprise AI adoption.
“LLMs give us a way to scale retail intuition,” Naresh says. “They capture the nuance behind human judgment and align it with the models that drive business outcomes.”
The intellectual rigor behind this work carries echoes of his scholarly contributions, including published research such as Delivering Actionable Insights: Combining Python, SQL, and Predictive Modeling Techniques for Customer Analytics and Dashboarding, where he explored how structured modeling frameworks can turn fragmented data into consistent intelligence.
Modern forecasting is not theoretical; it is measured in working capital, inventory turnover, and supply chain efficiency. Macy’s has publicly reported a 15 percent improvement in inventory turnover relative to 2019, supported by data driven forecasting and smarter replenishment. The company’s transformation plan outlines more than 100 million dollars in supply chain savings tied to advanced analytics, automation, and modernization initiatives.
The forecasting ecosystem Naresh architected is one of the engines behind these shifts. With long horizon accuracy, the company reduces over buys, prevents stockouts, and protects margin in fashion driven categories where volatility is highest. The LLM enabled scenario agent accelerates planning cycles dramatically, reducing analysis time from hours to seconds. These improvements ripple across merchandising, planning, logistics, and highly automated fulfillment centers such as Macy’s China Grove facility.
“Forecasting is becoming conversational,” Naresh reflects. “When a planner can ask the system a question and get a fully reasoned answer instantly, the entire business operates with more confidence.”
For Naresh, judge at Business Intelligence, leadership is as much about system predictability as it is about innovation. His signature is building engineering frameworks that are reproducible, governed, and easy for cross functional teams to adopt. He unified legacy systems into a consistent architectural model that supports analysts, planners, and downstream supply chain systems. This elevates organizational alignment and reduces technical friction inside large scale decision workflows.
His work earned formal recognition within Macy’s through a Spot Bonus, reflecting the program’s strategic importance and influence on the company’s modernization roadmap.
“Success is measured by what becomes easier for others,” Naresh says. “When people can make faster, better decisions, the architecture is doing its job.”
The next decade in retail will be defined by companies that treat AI as operational infrastructure rather than an experimental tool. Demand volatility, supply chain risk, and accelerating consumer cycles will make intelligent forecasting and AI agents essential. Naresh envisions an environment where LLM agents autonomously propose buys, evaluate tradeoffs, and reconcile cross functional priorities before plans ever reach a human reviewer.
The foundations for that future are already visible in the systems he helped build. His combined experience as an engineer, and published author has shaped a leadership style rooted in rigor, clarity, and enterprise scale impact.
“Forecasting is evolving into a partnership between human judgment and AI driven reasoning,” he says. “Our role as data leaders is to build the ground truth that partnership depends on.”
In an industry where precision is profit and infrastructure is strategy, Naresh represents the new generation of data leaders turning complex AI ecosystems into durable competitive advantage.