Kyan Cox

Incoming @ Coinbase | CS & Stats @ UW-Madison | Prev @ Gemini

Rye, New York, United States

About

Experience

  • Member of Technical Staff at Intelligible AI
    Dec 2025 - Present · 7 mos

    Early technical contributor on core engineering team; enabling LLMs to understand and use tabular data.

  • Incoming Software Engineer Intern at Coinbase
    Nov 2025 - Present · 8 mos

    Consumer Product Group; Beginning Summer 2026

  • Undergraduate Teaching Assistant at UW–Madison Computer Sciences
    Aug 2025 - Present · 11 mos

    Supporting CS 577: Introduction to Algorithms under Professors Marc Renault and Dieter van Melkebeek. ❖ Reduced quiz grading turnaround time by 59% for 700+ students by developing a Python CLI tool that parses Canvas quiz exports and generates student-specific PDF reports. ❖ Provide academic support through 10+ office hours weekly, answering individual questions and guiding students through problem-solving strategies. ❖ Host 14 weekly hour-long study sections for groups of 20+ students, reinforcing concepts such as Greedy Algorithms, Divide and Conquer, Dynamic Programming, Network Flow, and NP-Completeness.

  • Software Engineer Intern at Gemini
    Jun 2025 - Aug 2025 · 3 mos

  • Undergraduate Research Assistant at UW-Madison Department of Materials Science and Engineering
    Sep 2024 - May 2025 · 9 mos

    Worked on a UW-Madison research project applying machine learning to optimize donation campaign outcomes for a local food pantry, under Professors Benjamin Afflerbach and Dane Morgan as part of the Wisconsin Undergraduate Research in Data Science group (previously Informatic Skunkworks). ❖ Designed an end-to-end Python ML pipeline using Pandas, Scikit-Learn, and XGBoost that predicts individual donation amounts within ≈ $45 on average, explaining over 60% of donation behavior for a local food pantry. ❖ Achieved 3× feature expansion, growing a raw 5-column donation dataset to 15+ enriched fields (demographics, campaign context, seasonality) through Python web scraping, API integrations, and exploratory analysis. ❖ Trained a Random Forest classifier on 40+ engineered features, correctly flagging 71% of future repeat donors. ❖ Integrated U.S. Census APIs to enrich 7,000+ donor zip codes with median income and distance-to-pantry metrics. ❖ Occasionally volunteering at The River Food Pantry as part of my undergraduate research. Project Name: Optimizing Rate of Return from Donation Campaigns using ML Modeling and Feature Analysis