Portfolio
Some of my personal and team projects showcasing Machine Learning applications, real Agentic AI solutions, Data Science & Analytics tools, and more. Code repositories can be found on my GitHub.
An agentic AI app that turns raw customer feedback into structured product intelligence. Product managers, founders, and customer-facing teams are able to identify what customers are asking for, why it matters, and what to build next. Built using an MCP (Model Context Protocol) server architecture to enable AI agents to utilize the tools similarly to how a human uses them via the application's user interface.
A personal CRM for staying in touch with a professional network
A single-user personal CRM that turns relationship maintenance into a short daily workflow. I built it on Next.js and Supabase, with a deterministic follow-up engine, Gmail and LinkedIn messaging, and AI drafts that never send without review.
MCP Capable App for Synthesizing Real Customer Product Feedback
An AI-powered product intelligence app with MCP capabilities that turns raw customer feedback into prioritized roadmap insights, executive summaries, and agent-accessible product recommendations.
Machine Learning driven valuation of single family homes
An AI assisted valuation assistant that generates a prediction price of single family homes. The final XGBoost national model achieved a median absolute percentage error (MAPE) of 1.58%, outperforming Zillow's published Zestimate benchmark of 1.74%.
Supervised & Unsupervised ML Models to evaluate AQI prediction
Built Machine Learning models to evaluate whether AQI prediction is possible from wildfire and weather data alone.
Generative AI chatbot with multiple "personas" using Vector Stores
A Generative AI chatbot with multiple "personas" including Homer Simpson, Barbie, and Jesus Christ. I built the "personas" by generating vector stores in Python with LangChain and OpenAI, then host them in Weaviate.