My most recent significant technical achievement was a six-day hackathon where my partner and I built the “LEAD” project investigating embedding model similarities — my first self-lead end-to-end research project. We moved extremely quickly through research cycles, implementing ideas, collecting data, visualizing results, and iterating based on findings, all while working 10am-1am daily. We discovered embedding models aren’t linearly connectivity-aligned as hoped, with our linear “stitches” performing no better than mean estimators. Technically, I learned crucial research skills — running experiments immediately, implementing controls first, serializing large vector datasets, and writing organized throw-away notebooks. Socially, I learned to align goals up-front, distribute work without bottlenecking, and maintain emotional honesty when tensions arose. While our final analysis contained errors (which we intend to amend soon), I’m proud of the interactive publication we developed in such a compressed timeframe and the tremendous amount I learned about both technical execution and effective collaboration.