Build a reusable, AI-operable ecommerce scraping workflow | Fixed-price

I’m looking for a freelancer to build a reliable catalogue-extraction workflow that I, or an AI assistant acting for me, can operate repeatedly. Please distinguish reusable components from merchant-specific adapters and configuration. INPUTS - Merchant URL and target country/storefront. - Agreed category scope, exclusions and required product fields. - Run configuration, including output location and supported operating environment. - For a resumed run, the saved checkpoint from the previous run. Use Decathlon Singapore: https://decathlon.sg/ as a concrete example when describing your approach and quotation. State which merchants your fixed price covers and the cost of adding another merchant. REQUIRED OUTPUTS FROM EACH RUN 1. Structured CSV and JSON catalogue/product-detail datasets with product and variant identifiers, relationships, source URLs and capture timestamps. 2. All agreed fields available on source pages, including titles, descriptions, notes, specifications, prices/currencies, availability, variants, image galleries and ratings/reviews. State explicitly what is supported and any exclusions. 3. Accurate URLs for every image in each PDP gallery, preserving gallery order and product/variant associations where available. No image downloads required. Flag unavailable or temporary links. 4. A discovered-URL inventory and coverage report reconciling each URL to a successful result, failure or exclusion, with discovery sources and separate product/variant counts. 5. Full-dataset validation results for duplicates, missing fields, invalid values and broken relationships. 6. Run logs, saved checkpoints and a clear completion status. Partial or blocked runs must be labelled as such, with actionable reasons. 7. A separate exception report identifying missing/inaccessible data, affected records, reasons and any limits on proving completeness. REQUIRED PROJECT HANDOVER - Source code, configuration templates, dependency versions and installation instructions. - A step-by-step operating guide suitable for me or an AI assistant, including start, resume, validation and troubleshooting commands. - A repeatable verification process I can run, including independent coverage checks and instructions for comparing selected records and image links against timestamped source evidence. - Documentation of supported sites, limitations, access requirements, hosting/service costs and maintenance needs. - Source-code ownership/licence terms and proposed support/correction terms. ACCEPTANCE CRITERIA - Following the documentation, I or my AI assistant can set up and run the workflow in the agreed environment without undocumented manual steps. - A complete run on the agreed merchant produces the required datasets and reports, with every discovered URL accounted for. Unresolved failures are not counted as completed products. No silent omissions. - Coverage is reconciled against available category listings, pagination, sitemaps or other independent discovery sources, with discrepancies and uncertainty disclosed. - An intentional interruption followed by resume preserves completed work and produces no missing or unintended duplicate records compared with an uninterrupted run of the same test inputs. - An independent rerun succeeds using the documented configuration and process. Changes to live source data are distinguished from extraction defects. - Missing fields, invalid data and access failures are detected and reported. Retries are bounded; blocked or partial runs never report full success. - Independently selected samples match captured source evidence. Sample size and accuracy thresholds will be agreed before hiring, and identified defects corrected before acceptance. Sample checks provide confidence, not proof that every field is correct. - Delivered code and documentation cover the agreed scope. A demonstration on a few products is not sufficient evidence of full-catalogue operation. PLEASE INCLUDE IN YOUR PROPOSAL - Relevant systems you personally built and maintained. - Recommended architecture and hosting, including whether a local computer must remain on. - Fixed build price, included merchant coverage, turnaround and milestones. - Recurring service costs, maintenance charges, exclusions and post-delivery support. Please quote for the reusable system separately from one-off data extraction. Clear English communication is required. PREFERRED TURNAROUND Within 2 weeks from the agreed start. Please quote your exact delivery time and flag if you cannot meet this preference.