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.