PDF Classification & Routing AI Worker

We are looking for a developer to build a **small local PDF classification and routing worker**. The worker will run on an existing client server. The input and output folders are already prepared. ### What it should do When a new PDF appears in an input folder: * Detect the new PDF * Send the PDF to a vision-capable AI model * Extract basic information such as customer, date, document type and document number * Classify the document into one of two business areas * Rename the PDF * Move it into the correct existing customer/year folder * Log the result If the document cannot be classified safely, it should be moved to a **manual review folder** instead of guessing. The PDFs can include normal digital PDFs as well as scanned documents, handwriting, tables or slightly messy scans. There is **no need to build a complex OCR system**. The PDFs can be sent directly to a vision-capable AI model. ### Important The existing folder structure already exists. The worker should: * never create new customer folders * never overwrite existing files * never delete PDFs * use manual review when unsure * run continuously in the background * include a simple dry-run mode for testing before going live No dashboard or large application is needed. This is a **small local automation project**, not a new platform. ### When applying Please send: 1. A short example of a similar automation or document-processing project 2. Your fixed-price estimate 3. Your estimated timeline 4. Which AI model/API you would use for PDF understanding