I worked with the MIT's Heavy Ion Group at CERN. Our group analyzed data at the CMS detector. However, heavy ion data (lead, specifically) was unavailable during my time with the team.
We hypothesized that we could would observe heavy ion-like behavior in p-p collisions due to the record-breaking energies at the LHC (7TeV at the time). Our group discovered many indicators of quark-gluon plasma in high multiplicity p-p events, characterized by non-uniform angular correlations.
A major concern about our study is that "pile-up" events (2+ distinct vertices mistakenly traced back as a single vertex) bias the statistical analyses. My task was to quantify the effects of pile-up on our study.
Analyzing billions of collisions (using C++ based "Root"), and comparing them to monte-carlo generated Pythia data, I determined that, while pile-up effects increase at higher multiplicity events (up to 40%), their effects could be distinguished with more advanced algorithms to bring indistinguishable pile-up rates down to 10%. This rate turned out to be tolerable to the rest of the study, confirming that the heavy ion-like correlations were not due to pile-up events.