We are looking for an experienced backend developer to design and build a scalable, data driven product matching system for our e-commerce operations. The goal is to create a system that automatically processes supplier product lists in different formats and matches them to our internal product catalog with high accuracy. This system must be configurable, maintainable, and future-proof, so that new suppliers and new product categories can be added without changing core code. Project Overview We work with many suppliers who provide product lists in (Excel/CSV). These lists need to be: - Matched to our internal catalog - Stored with full traceability - Prepared for internal pricing and stock updates You will build a backend system that handles this end-to-end process. Key Responsibilities - Design and develop a Python based matching engine - Build a flexible supplier ingestion system driven by configuration (not hardcoded rules) - Integrate with Python back end of our ingestion web application - Implement intelligent matching logic structured title parsing - Build category-based matching profiles (phones, watches, accessories, etc.) - Implement controlled AI usage for specific cases (e.g., color interpretation) - Ensure full logging, confidence scoring, and explainability of matches - Create a system that supports manual overrides and future improvements Core Requirements 1. Universal & Configurable Architecture The system must be designed so that: - New suppliers can be added via configuration - Column mappings can be adjusted without code changes - New product categories (e.g. toothbrushes, microwaves, or other appliances) can be added through configuration and database profiles - Core matching logic remains unchanged - Adding new products or categories must be mainly a configuration task, not a development task. 2. Supplier File Processing - Support Excel and CSV files - Store supplier configurations in the database - Handle frequent format changes without redeployment 3. Matching Engine - Use structured parsing of titles and attributes - Extract features such as brand, model, storage, size, color, etc. - Score multiple candidates and select the best match - Provide confidence scores and explanations 4. Database & Infrastructure - Azure SQL Server - Central storage for: - Product catalog - Supplier configurations - Matching profiles - Processing runs - Results and audit logs Technical Stack (Preferred) - Backend: Python - Database: Azure SQL Server - APIs: REST-based architecture - AI: Azure OpenAI (limited, controlled usage) - File Processing: Pandas / OpenPyXL / similar Required Skills - Python backend development - Experience with data processing pipelines - SQL (preferably SQL Server / Azure SQL) - Experience with system architecture and scalability - API design - Data normalization and matching logic Nice to Have - Experience with AI-assisted workflows - Knowledge of product catalog systems - Experience building configurable tools What We Value Most We are not looking for a quick script. We are looking for a developer who can: - Think in systems - Design for long-term scalability - Build maintainable architecture - Document decisions - Deliver production-ready software This project is intended to be a long-term foundation for our operations. Project Scope - Initial system design - Core backend development - Database schema design - Matching engine implementation - Configuration framework - Documentation - Support during testing phase Potential long-term collaboration for maintenance and extensions. Additional Information - The ZIP file includes documentation and explanations of our current database structure and data model. Access to live environments and production systems will be provided in a later project phase, once the collaboration is established. - The ZIP file also contains an early prototype of the matching engine. This prototype is incomplete and not production-ready. It was created during an exploratory phase to better understand the problem and define the required scope. It is provided for context only and should not be considered a technical foundation for the final solution.