Product Manager, Red Team, Platform Responsibility

Team Introduction The team is responsible for testing and evaluating global content security systems from an attacker's perspective, identifying potential vulnerabilities, and providing improvement recommendations, ensuring content safety across cultural and linguistic contexts.

Responsibilities

• Threat Discovery & Risk Analysis: Lead structured threat pattern discovery from multiple signal sources (TikTok online, external escalations, competitive intelligence, threat intelligence). Identify emerging abuse patterns and evasion techniques; maintain and expand the team's adversarial knowledge base (Evasion Taxonomy, Abuse Pattern framework). Translate raw signals into structured, actionable threat pattern cards. - Attack Sample Construction: build the end-to-end sample construction pipeline - from threat pattern intake to high-fidelity adversarial sample production at scale. Define construction methodologies, tooling requirements, and quality standards. Ensure sample coverage systematically maps to the Abuse × Evasion attack surface. - Adversarial Evaluation & Remediation: Design and execute adversarial evaluation programs that measure the platform's true defensive capability under realistic attack scenarios. Establish evaluation cadences, coverage metrics, and reporting mechanisms. Drive the feedback loop between Red Team findings and downstream policy, model, and operations improvements. - AI Agent Development: Lead the design and deployment of LLM-powered agent systems to automate and scale Red Team operations, including signal triage, evasion pattern matching, abuse pattern analysis, and sample construction guidance. Define agent workflows, human-in-the-loop review protocols, and automation quality benchmarks. Continuously evaluate and expand the boundaries of AI-driven adversarial analysis.