I summarized my career vision and why I joined the externship, linking my analytics skills to a goal of working on data and predictive modeling in healthcare.

I created a stakeholder map and role summary for a nephrology clinic, documenting responsibilities, workflows, and pain points across MDs, RNs, APPs, and PCCs.

I mapped clinic patient and internal workflow challenges and prioritized critical escalation failures for deeper analysis.

I created a text-based decision tree for outpatient hyperkalemia escalation, translating clinical criteria into stepwise triage and escalation rules.

I selected and evaluated public health datasets, then chose MEPS as the best source to study how psychological distress affects cancer care in underserved U.S. populations.

I collected, cleaned, and merged county-level SNAP and colorectal cancer data, then ran quality checks and basic transformations to prepare a merged dataset for analysis.

I performed exploratory data analysis and produced visualizations and statistical tests on county-level cancer and socioeconomic data.

I drafted a county-level research report linking socioeconomic and demographic indicators to colorectal cancer risk in Iowa (2016–2020).

I analyzed a mortgage document package to identify missing and extraneous items, and described how AI could automate detection and extraction across varied formats.

I cleaned and standardized messy loan and text data, handled encoding issues, and prepared images for OCR using Python in Google Colab.

I extracted text and bounding boxes from a mortgage loan worksheet using PyMuPDF, then reconstructed tabular fee data into structured JSON.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned mortgage PDF and reported which engines produced the most usable text and bounding boxes.

I built an end-to-end RAG pipeline using LlamaIndex, creating document parsing, semantic chunking, embeddings, and retrieval to power an LLM-backed Q&A workflow.

I tested and compared embedding models for a mortgage-document RAG pipeline and selected the best fit for CPU-bound workflows.

I built a page-level segmentation and classification pipeline that tagged pages and grouped them into documents with metadata.

I built a Gradio-based PDF Q&A chatbot interface that accepted uploads, displayed status, and let users save chat history.

I built and tested a Moodboard Generator agent in n8n, iterating prompts and workflow nodes until the agent produced image-ready prompts and partial visuals.

I built a multi-source Trend Discovery Agent that consolidated RSS, Google, and Amazon data into a single trend report.

I built a Competitor Monitoring AI Agent that merged scraped Wayfair, Amazon, and Ruggable data and produced business-facing analysis under multiple system-message lenses.

I refined an AI Insights & Content Agent to turn scraped product and competitor signals into brand-aligned marketing content.

I drafted an evidence-based briefing note linking frontline attrition signals to operational costs and suggested next investigative steps.

I cleaned and structured Glassdoor and YouTube employee feedback into analysis-ready datasets, documenting cleaning choices and data issues.

I cleaned and preprocessed noisy employee feedback text to produce analysis-ready text columns for keyword and sentiment work.

I identified three themes in employee feedback, quantified their sentiment, and investigated the top theme with a root-cause analysis.

I generated a ranked list of the top pain points for wireless headphones using AI-sourced review excerpts and frequency estimates.

I built an AIDA-based customer journey map for premium headphones, mapping awareness through action with customer thoughts, feelings, and behaviors at each stage.

I analyzed survey responses to identify consumer needs and segment behaviors, combining quantitative charts and qualitative coding to surface purchase drivers and product gaps.

I delivered an Amazon reviews analysis comparing Beats by Dre and competitors, combining EDA and TextBlob sentiment measures into a concise report.

I documented the database, wrote SQL queries, and captured quick answers to the team's recurring questions in a living Week‑1 playbook.

I designed and built a summary table and reorder recommendation from joined product and sales data.

I built a Q4 ad allocation plan connecting Meta campaign spend to revenue and recommended which campaigns to keep, cut, or scale.

I wrote a diagnosis report that identified bot-driven noise, recalculated true cart abandonment, and proposed actionable fixes for Breaking Games.