I'm Peter Xenopoulos. I'm a research scientist at NVIDIA, focusing on developing machine learning and visual analytics solutions for esports & sports. Previously, I led quantitative research for a sports betting exchange, worked with a professional baseball team (Phillies), analyzed companies for a top seed-stage venture capital fund (Tuesday VC), created data and modeling pipelines for an athlete investment fund (Big League Advance), and worked on the AI data science team at Facebook. I maintain the following projects:
- π©βπ» awpy, a library to parse, analyze and visualize Counter-Strike demofiles. The project has spawned published papers in IEEE Big Data, ACM's WWW conference, and IJCAI.
- πΎ ESTA, a large esports dataset, which contains 8.6m player actions, 7.9m game frames and 417k player trajectories from 1.5k professional CSGO matches.
- π PyCalibrate, a visual analytics tool to analyze model calibration. This work was accepted to IEEE VIS 2022.
- π‘ GALE, a technique to measure the similarity between sets of local explanations developed in collaboration with Capital One. Accepted for a spotlight talk at the TAGML workshop at ICML 2022.