Ryan Cecil

Hi! I am a PhD candidate in Statistics at the University of Pittsburgh. My thesis develops rigorous methods for comparing groups of well-performing predictive models — often called model confidence or Rashomon sets — and drawing valid conclusions from them. My advisor, Lucas Mentch, and I address this comparison problem through our recent work on Model Class Selection. I have also developed and applied statistical and machine learning methods in the fields of computer vision, population genetics, and criminology.

Beyond academia, I have worked with data science based start-ups in the Pittsburgh region. As an intern at Behaivior, a digital health company, I developed machine learning models to detect craving events in those recovering from opioid addiction. More recently, over the past few summers, I have worked closely with the Founder and CEO of Allref, a life sciences AI company, to build the underlying language graph models powering their platform.