I chose and committed a county-level CDC dataset linking food insecurity and diagnosed diabetes for correlational analysis.

I mapped the CDC PLACES county dataset and documented its structure, fields, and preprocessing steps to prepare two measures for analysis.

I ran a Pearson correlation on county-level data linking food-insecurity prevalence and diagnosed-diabetes prevalence.

I stress-tested my county-level finding on food insecurity and diagnosed diabetes and produced a documented checklist with limits and a revised interpretation.

I analyzed a pharmaceutical sample file to identify document types, date formats, and challenges AI faces when extracting structured data.

I cleaned a messy pharmaceutical dataset using Pandas, fixed format inconsistencies and invalid entries, and exported a verified CSV.

I extracted word-level text and bounding boxes from a 3-page SDF, then grouped and structured those tokens into searchable records and field candidates.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned pharmaceutical Certificate of Quality and produced a layout-aware recommendation report.

I built and tested a complete RAG pipeline for document retrieval using LlamaIndex and hybrid retrieval methods.

I ran three RAG configurations on a sample SDF, compared outputs, and identified the best retrieval setup.

I built a page-level segmentation and metadata output pipeline for bundled pharmaceutical PDFs, producing structured JSON that labeled document boundaries and types.

I built a Gradio chatbot interface that connected to a local retrieval pipeline, added multi-PDF upload, selectable answer modes, source/page references, processing feedback, and chat export.

I created a stakeholder map and role summary for a case manager in mental healthcare, then shared the map link.

I mapped patient-facing issues and internal process gaps that lead to missed post-discharge follow-up and selected the highest-priority problem for root-cause analysis.