I audited a mortgage document package and identified missing items and processing issues, then outlined how AI could automate classification and validation.

I prepared and processed mortgage-related documents using Python, applying data cleaning, JSON handling, text standardization, and image preprocessing to improve OCR readiness.

I extracted and structured key fields from multi-page mortgage PDFs using Python and PyMuPDF, producing word-level text with bounding boxes and heuristics to locate loan fields and fees.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned mortgage PDF and produced a brief report summarizing accuracy, output structure, and usability for downstream AI tasks.

I built and evaluated a Retrieval-Augmented Generation pipeline that indexed a PDF, implemented query fusion retrieval, and generated answers from retrieved content.

I tested and compared embedding models for a RAG pipeline and selected the best performer based on retrieval scores and qualitative checks.

I built a page-level metadata extractor and document-boundary classifier for multi-document PDFs, producing structured JSON that mapped pages to document IDs and types.

I built a Gradio chatbot UI that accepts a PDF upload and answers questions by querying a local RAG pipeline.