Portfolio

I learn best by building. Here's what curiosity and a love for data have led me to create.

Predicting Hourly Bike Rentals
Machine Learning

Predicting Hourly Bike Rentals

In Pursuit of Model Accuracy!

A data science project to predict hourly bike rentals in the city of Seoul using date and weather seasonal data.

Rtidyversetidymodels
Customer Segmentation & Persona Creation
Analytics

Customer Segmentation & Persona Creation

A Case Study Using K-Means Algorithm

This project uses k-means clustering to segment banking customers and create targeted personas for improved engagement and personalized financial product offerings.

Rtidyversetidymodels+1
Machine Learning for Email Lead Scoring
Machine Learning

Machine Learning for Email Lead Scoring

A Case Study of Predictive Analytics in Marketing

Building a machine learning model to score email leads based on conversion likelihood, connecting predictions to ROI and deploying with FastAPI and Streamlit.

PythonPandasScikit-Learn+2
CRM Data Analysis
Analytics

CRM Data Analysis

Understanding and Improving Sales and Marketing Strategies

Analyzing a year's worth of B2B sales pipeline data to establish benchmarks, identify seasonal patterns, and evaluate sales rep performance.

Rtidyverseggplot
Sales Pipeline Analytics AI Assistant
AI

Sales Pipeline Analytics AI Assistant

Conversational Analytics for B2B Sales Pipelines

An AI assistant that lets stakeholders ask plain-English questions about sales pipeline performance and get back charts, tables, and summaries in real time.

AILangGraph
PythonLangGraphOpenAI+3
Cannondale Bikes AI Assistant
AI

Cannondale Bikes AI Assistant

An Agentic RAG System with Tool-Calling and Conversational Memory

An AI agent that acts like a knowledgeable bike shop employee, using five specialized tools to search, summarize, compare, and recommend Cannondale bikes through natural conversation.

AIRAG
PythonLangChainOpenAI+2
Building My First "Useful" Automation with Claude Code
Automation

Building My First "Useful" Automation with Claude Code

A weekly scraper, a Next.js dashboard, and a Google Calendar integration, end to end.

How I solved a personal pain point — tracking DC-area data and AI meetups — by building an end-to-end automation with Claude Code, a Next.js dashboard, and Google Calendar integration.

AutomationClaude Code
Claude CodeNext.jsTypeScript+2