I analyzed a mortgage file to find which documents were missing and assessed document relevance and processing challenges.

I prepared and cleaned mortgage-related data and images using Python in Google Colab, applying Pandas and OpenCV/PIL techniques.

I extracted and structured text from a mortgage PDF using PyMuPDF, then proposed layout-aware grouping to convert tokenized output into clean fields.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned mortgage PDF and reported which engine worked best for my workflow.

I built and tested chunking and embeddings in a LlamaIndex Colab, troubleshooting library issues and API integration until the pipeline ran.

I tested and compared embedding models for a RAG pipeline and chose the best performer.

I built a page-level segmentation and metadata pipeline using an open-source embedding model and rule checks to group pages into documents.

I built a Gradio chatbot that let users upload a PDF and ask questions, connecting document uploads to a retrieval pipeline and returning grounded answers in a chat UI.

I mapped the workforce roles and responsibilities inside an Amazon Fulfillment Center, detailing who does what, who they work with, and the tools they use.

I cleaned and structured employee feedback from Glassdoor and YouTube, producing analysis-ready datasets and examples of strong and weak employee voice.

I cleaned a sample of employee feedback text for analysis, normalizing case, removing punctuation, and filtering stopwords while preserving certain contractions.

I produced a root cause analysis and recommendations for the priority theme, using the 5 Whys to trace scheduling problems back to planning and incentive structures.