Hello there, I'm Marco 👋

• Assessed and optimized AI-generated code for improved reasoning and output quality.
• Created coding challenges in Python, HTML, Java, and SQL to enhance model capabilities.

• Analyzed USL soccer league data using Poisson distribution to predict match outcomes.
• Built custom Python scripts to parse and process data for 450+ matches.
• Came out with 50% accuracy in predicting match results, demonstrating the model's effectiveness.

• Worked with Active Directory and JUnit over SSH on Linux servers.
• Used Splunk and Kibana to monitor and report suspicious account activity.
• Maintained organization of sensitive customer data.

• Conducted funded research on superhydrophobic surface design about micro/nano-structured coatings.
• Presented findings at the NCUIRE Symposium at Arizona State University. Awarded 1st Place (non-ASU category) and 3rd Place (overall).
• Collaborated with faculty mentor to design experiments and interpret results.

Created in a 48-hour hackathon platform for the Arizona underground rave scene. Claude vision API auto-parses uploaded event flyers into structured data, with all AI calls proxied through Firebase Cloud Functions to keep API keys server-side. Features real-time Firestore social feed, Mapbox geocoding with Firestore-cached venue data, and an AI design chatbot for flyer creation guidance.

Hackathon built analytics dashboard for Polymarket MCP server integration, leveraging Claude AI for insights with contextual related chart data, NewsAPI for related articles, and recent trade data with filtering for real-time updates.

Multi-role forum platform with role-based access, moderation, and persistent data storage.

Real-time flight data dashboard used at an airport retail kiosk to monitor schedules.