Python Developer — Automated PDF Data Extraction from Architectural Plans to Structured Output

Description: We are building an automation pipeline for a Swiss construction-tech product (Bausoft) that ingests architectural plan PDFs and extracts the final results into a clean, structured format ready for downstream processing. The PDFs are a mix of vector-based and scanned architectural/bathroom floor plans, mostly in German. The pipeline must reliably read each document, extract the relevant data — room labels, dimensions and measurements, fixture annotations, title-block metadata, and tabular values — and output the results in a defined structured format (JSON schema and/or Excel) with consistent field naming and units. Scope of work: PDF ingestion module handling both native-text and scanned (OCR) documents Text and table extraction with German-language support, including abbreviation handling Layout-aware parsing to associate values with the correct rooms/elements Structured output generation (JSON per our schema, plus Excel export) Validation layer flagging low-confidence extractions for manual review Documentation and a simple CLI or API endpoint to run the pipeline Required skills: Python, PyMuPDF or pdfplumber, Tesseract or a comparable OCR engine, pandas, JSON schema design. Experience with architectural/CAD drawings, German documents, or LLM-assisted extraction (e.g., Claude/GPT structured outputs) is a strong plus. Engagement: Fixed-price, milestone-based. Please share relevant PDF-extraction samples and briefly describe how you'd handle a scanned plan where OCR confidence is low. Start your proposal with the word "Blueprint" so we know you read the full post.