Abstract: Smart retail—having in-store and online components—permits companies to provide support services that complement (but do not duplicate) customer value and convenience. It is thus feasible to address inventory optimization as an AI problem, taking into consideration customer needs for timely product availability without long delivery lead times. Data engineering principles reformulate inventory optimization as a recommendation engine, predicting future warehouse, store, and web inventory levels in the short and medium term for long-lead-time products to assist decisions on how much to order. A concept for a fully receptive data architecture is introduced, capable of supplying the large amount of quality-cleaned data required to train the AI models and to implement AI-based data pipelines that spatially distribute Web inventory recommendation across the supply chain. These pipelines, optimized for fast local machine learning (ML) workloads, reduce the volume of data sent to the core DB and the number of jobs initiated there, thus accelerating inventory-level refresh by making large amounts of inventory-ready data locally available.
The data architecture, supporting the full data lifecycle in accordance with the smart retail concept, consists of data-ops pipelines designed for fully receptive external and internal data flows and data-engineering lines for preparatory and loading jobs dedicated to core BI information. An additional component dedicated to the implementation of AI-based data pipelines is sized to cope with the spatiotemporal distribution throughout the modelled area of slow-loading-tagged external data. Inventory-level refresh is accelerated by minimizing the volume of data sent to the core DB and the number of jobs initiated there, thus enabling core data availability that supports fast local ML workloads and local supply-demand analysis.

Keywords: Smart Retail Systems, Omnichannel Retail Architecture, AI-Based Inventory Optimization, Inventory Recommendation Engines, Predictive Inventory Forecasting, Retail Data Engineering, Smart Supply Chain Analytics, Web And Store Inventory Integration, Data-Ops Pipelines In Retail, AI-Driven Data Pipelines, Local Machine Learning Workloads, Spatiotemporal Inventory Distribution, Retail Data Lifecycle Management, Inventory-Level Refresh Optimization, Edge-Optimized Retail Analytics, Demand–Supply Alignment, Long Lead-Time Product Planning, Retail BI Data Architecture, Scalable Retail Data Platforms, AI-Enabled Inventory Decision Support.

Downloads: | DOI: 10.17148/ IJIREEICE.2025.131228