I want to enrich my online shoe store with an AI-powered recommendation engine that studies each shopper’s purchase history and instantly serves up the pairs they are most likely to buy next. The model can draw on three data streams—user account data, my e-commerce platform records, and any third-party customer datasets I supply—to build a unified profile and surface truly personal suggestions. Here is what the finished job looks like from my side: • A trained model (Python preferred, TensorFlow or PyTorch are both fine) that ingests the above data sources, updates itself regularly, and outputs ranked product recommendations in real time. • An API or embeddable snippet I can drop into the product and home pages to display “You might also like” shoes, along with a lightweight admin panel where I can adjust thresholds and view basic analytics (CTR, conversion uplift). • Clear documentation covering data-prep steps, model retraining, and integration hooks so my dev team can maintain it long term. • A short hand-off session (recorded) walking me through deployment and future tuning. If there’s anything you need—schema samples, anonymized historical orders, or storefront staging access—just ask and I’ll provide it right away.