Applied Data & Technical Projects
Independent projects built to keep my technical skills sharp alongside my implementation and customer success work — automation tools, dashboards, and ML applications built with Google Apps Script, SQL, Python, and modern web tools.
Automation & Implementation Projects
Built directly out of the implementation and customer success work I do every day — automating the manual reporting and cleanup tasks that shouldn't still be manual.
KPI Command Center
A Google Apps Script KPI dashboard and reporting system that turns raw support case data into an executive dashboard and team performance view — SLA tracking, response/resolution metrics, and auto-generated charts, built entirely in Google Sheets.
Workflow Automation & Task Tracker
A Google Sheets task tracker with a custom menu for one-click updates, automatic dashboard refreshes, completed-task archiving, and scheduled daily refresh triggers via Apps Script.
Sales & Lead Funnel Dashboard
A Google Sheets tool that validates a raw CSV lead export before building the dashboard — conversion rate, win/loss, deal value, and owner performance, refreshed automatically.
Spreadsheet Rescue & Optimization
A safety-first Apps Script add-on that audits a messy spreadsheet, generates a plain-language health report, and backs up the file automatically before any cleanup runs.
Implementation Tracker
A full-stack project management tool for implementation managers — track projects, milestones, and risks, with an auto-computed health score and status reports exportable as PDF. Built with React, FastAPI, and SQLite.
Applied AI Projects
LLM-powered tools — retrieval, grounded generation, and cited answers built on the Claude API.
PyIndex — AI-Powered Enterprise Search
A Glean-style enterprise search demo built with FastAPI and React — ingests real documentation from the PyPI API, indexes it with TF-IDF search, and generates Claude-powered, cited answers on top of the raw results.
Pathlight Onboarding Assistant
A grounded AI chatbot for a fictional K-12 platform — ask a navigation question in plain English and get back the exact click path, cited to the source guide entry. Built with Flask, Claude, and a custom retrieval layer to pair instructional design with applied AI engineering.
Clinical Text Extractor
A Streamlit dashboard that transforms unstructured clinical notes into structured, analyzable data - highlighting my skills in API integration, LLM-powered extraction, SQLite logging, and interactive visualizations to help users identify diagnoses, prescriptions, and follow-up plans with clarity and ease.
Data & Machine Learning
Earlier projects built to sharpen my SQL, Python, and machine learning skills.
Personal Knowledge Base Search
A FastAPI + Streamlit search tool over a Wikipedia-derived corpus that ranks results two ways — TF-IDF keyword matching and LSA semantic matching — with a side-by-side comparison view. No LLM involved.
Creator Analytics Dashboard
A Streamlit dashboard powered by the YouTube Data API v3 that compares up to four YouTube channels side by side — subscriber counts, upload cadence, engagement rate, and top-performing videos, visualized with Pandas and Plotly.
Term Deposit Subscription Predictor
Built a full-stack machine learning app using Random Forest and SMOTE to predict whether a banking customer will subscribe to a term deposit. Includes interactive dashboards for model evaluation, batch prediction, and customer segmentation via Streamlit.
Customer Churn Explainability Dashboard
An interactive Streamlit dashboard that uses SHAP (SHapley Additive exPlanations) to interpret an XGBoost churn model — upload a dataset, train the model, and explore predictions through summary, scatter, waterfall, and decision plots.
Customer Health Dashboard
Built a SQL-driven account health scoring engine — window functions and CTEs over synthetic login, feature-adoption, and support-ticket data — to produce a 0–100 health score and risk tier per account. Surfaced through both a Streamlit app and a zero-install HTML dashboard with a live, filterable demo.
Live Data Stream Dashboard
A real-time dashboard built with FastAPI, WebSockets, and React that streams live clickstream data and purchase activity. Includes dynamic Chart.js visualizations, CSV export, and logs live product interactions. Built for performance monitoring and behavioral analysis.