AI Product Recommendation Website

I plan to launch an e-commerce site whose standout feature is an AI-powered product recommendation engine. The model should study live user-behaviour signals—click paths, dwell time, search queries—and blend them with product metadata (titles, tags, attributes) to surface highly relevant items in real time. A streamlined customer journey is the priority, so latency has to stay low and the suggestions must update dynamically as shoppers browse. Whether you prefer a custom Python stack with TensorFlow or a managed service such as AWS Personalize is up to you, as long as the pipeline remains transparent and can be fine-tuned once enough behavioural data accumulates. Deliverables • Responsive e-commerce front-end connected to the recommendation API • Data-collection layer that captures user actions without slowing the site • Training and inference scripts (or configuration files) for the recommendation model • Simple admin panel to monitor key metrics and manually trigger re-training • Deployment instructions so I can reproduce the setup on a fresh server or cloud account Acceptance criteria 1. New users see cold-start recommendations based on product metadata. 2. Returning users receive personalised suggestions that reflect at least their last five interactions. 3. End-to-end page load plus recommendation refresh stays under two seconds on a standard broadband connection. If you can weave this engine seamlessly into a clean shopping experience, let’s get started.