Employer description & source details
Own the enforcement lifecycle. Lead a defined abuse or enforcement problem from problem framing and signal discovery through labels and taxonomies, model or rule development, thresholding, rollout, monitoring, and iteration. Discover and characterize abuse. Analyze large-scale behavioral, content, account, and network data to identify adversarial patterns, emerging threats, and gaps in existing mitigations. Build and evaluate decision systems . Develop models, classifiers, scoring systems, rules, and analytical prototypes using the statistical, machine-learning, or GenAI methods best suited to the problem. Define decision-quality standards. Create evaluation frameworks that address precision, recall, calibration, coverage, robustness, fairness, user friction, reviewer capacity, and other operational costs—not model performance in isolation. Ship with engineering partners. Partner with Engineering and Safety ML to productionize promising approaches; define data and logging needs, launch guardrails, monitoring and alerting, escalation paths, and rollback criteria. Translate policy into systems. Work with Policy and Safety Operations to convert written policy and product requirements into modelable targets, labels, decision rules, reviewer workflows, and measurable enforcement outcomes. Measure real-world impact. Design staged rollouts, experiments, and observational analyses that test whether interventions reduce harm while managing false positives and downstream effects on users, communities, and operations. Respond and learn. Support investigations and urgent escalations, identify mitigation options, conduct post-incident analyses, and turn one-time findings into durable detection or monitoring improvements. Scale the team’s effectiveness. Build reusable datasets, evalu
Extracted skills: python, sql, machine learning, artificial intelligence, cybersecurity, security, operations