Content-Based Electronics Recommender

I’m building a shop that sells only electronics and I want to add a content-based recommendation engine so every visitor sees items that match what they’re already browsing. The model must rely on product attributes—specifically brand and the key technical specifications we hold in our catalog—rather than purchasing history or user ratings.

You’ll receive a clean CSV export with the full product feed (titles, descriptions, brand, and spec fields such as screen size, storage, chipset, etc.). Your task is to:
• engineer meaningful feature vectors from those brand and spec columns,
• train and evaluate a content-based model that returns the top-N similar products for any given item ID, and
• hand back well-commented code (Python is fine, but I’m open) plus a short README that shows how to retrain and query the engine.

Acceptance is simple: point the script or API at the dataset, call it with a sample product, and the response must list sensible alternatives that clearly share the same brand or overlapping specifications. If you’ve built similar recommenders before, let me know—otherwise I mainly care that the logic is transparent and straightforward to extend later.