I need a basic decision-support system that helps me decide what jewellery stock to buy and when to buy it. The core of the job is to connect three things: • Image metadata that will be extracted from product photos. • A cosine-similarity comparison that measures how well a buyer’s text query matches each item’s metadata. • Historical signals such as purchase date, sale date, and how long items have been sitting in inventory. For now, I only have the image metadata in hand, so please set up a simple pipeline—Python is fine—to pull colour, shape, or texture details from new images and store them in a table I can expand later with transaction and inventory records. On top of that data model, build an initial Power BI dashboard where I can: 1. Enter or edit stock details directly (SKU, category, cost, selling price, dates). 2. See current stock on hand and ageing by category. 3. Spot which ornaments deliver the highest profit and which ones tie up cash. 4. Get a quick recommendation list based on the similarity scores and historic performance. Deliverables: • Clean, well-commented code that extracts metadata and runs the cosine-similarity search. • A lightweight data model (CSV or Excel is fine) linking image, inventory, and transaction tables. • A Power BI file (.pbix) with interactive visuals and slicers for the points above. • Step-by-step setup notes so I can refresh the data locally without extra licences. This first version can be a straightforward prototype—keep it simple, functional, and easy to extend in future phases.