{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/f40e6a858c684b91973baff2f7c30e11\" frameborder=\"0\" width=\"1662\" height=\"1246\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1246,"width":1662,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1246,"thumbnail_width":1662,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/f40e6a858c684b91973baff2f7c30e11-fb6a1ec3363a51b4.gif","duration":656.55,"title":"X Bot Detector Recorded Demo (Mohamed Bucheeri)","description":"Hi everyone, I built a Chrome extension called XBotDetector to score X profiles as human, bot, or uncertain, and show which features drove the score. I trained three models on 37,438 labeled accounts, with an XGBoost model using 37 numeric features that achieved an F1 of 0.8076 and trained in about one second. The feature engineering in data.py is the most important part, and explainability comes from per-feature contributions from the XGBoost model. The backend runs with FastAPI locally, and no data leaves the browser. No action was requested from viewers."}