Retail Shelf Intelligence Platform

It is an enterprise-grade SaaS platform for retail shelf analytics, product recognition, and in-store execution monitoring. It enables FMCG brands and retailers to capture shelf photos during store visits, automatically detect products using AI-powered image recognition, and evaluate store compliance through configurable KPI rules.

Key Features

• AI-Powered Shelf Recognition — Field representatives photograph store shelves; the system stitches panoramic images and runs multiple recognition engines to detect and classify every product on the shelf, including empty spaces and price tags.
• KPI Engine & Scorecards — Configurable rule engine evaluating Share of Space, Share of Facing, Product Block compliance, Proximity analysis, and Group Presence per visit. Results are aggregated into scorecards segmented by store groups.
• Product Catalog Management — Hierarchical SKU management across Master → Domain → Country → Store levels, with versioning, multipack support, barcode tracking, and multi-language descriptions.
• Store & Display Management — Stores organized by geography hierarchy, categories, and groups. Displays and modules define physical shelf layouts via planograms with container-level granularity.
• Survey Builder — Dynamic in-store surveys built with SurveyJS, assignable to specific stores, with response tracking per visit.
• Data Collection & Annotation — Fabric.js-based annotation tool for manual labeling and verification of AI recognition results, with full event audit logging.
• Multi-Tenant Architecture — Domain-based data isolation supporting multiple clients, countries, and languages within a single deployment.
• Role-Based Access Control — Granular claim-based authorization with 30+ permission types controlling access at route, component, and API endpoint levels.

  Tech Stack

• Frontend: Angular 17, PrimeNG, Angular Material, Bootstrap 5, Fabric.js, FullCalendar, SurveyJS
• Backend: ASP.NET Core 7 (Clean Architecture), MediatR (CQRS), Autofac, FluentValidation, Hangfire
• Database: PostgreSQL (143 entities, EF Core Code-First)
• Auth: JWT with refresh tokens, claim-based authorization
• Monitoring: Serilog, Graylog
• CI/CD: Azure Pipelines

• Architecture Highlights

• Clean Architecture with separated Presentation, Business, Data Access, Core, and Entity layers
• CQRS pattern via MediatR with command/query handlers and validation pipelines
• Background job scheduling (Hangfire) for shelf share recalculation, image stitching cleanup, and visit auto-completion
• Redis caching layer with aspect-oriented cache invalidation
• SignalR for real-time updates
• Multi-language support (English, Turkish, Spanish, Catalan, Arabic) with RTL handling
• API versioning (v2) with Swagger/OpenAPI documentation