CareerPlanGet AI match score →

Threat Intel Manager, Model Exploitation & Fraud

San Francisco, CA💼 Full-time💰 $375,000–$375,000🗓 2026-07-08 → 2026-07-31

Core

Build and run the Model Exploitation & Fraud team to detect, investigate, and disrupt large-scale exploitation of AI systems, including model distillation, unauthorized access, account farming, and fraud operations.

Role type

Strategy-level Threat Intel Manager (Model Exploitation & Fraud)

Builds

Detection systems, investigation playbooks, and external intelligence sharing programs for AI safety and platform integrity.

Domain

AI Safety / Threat Intelligence / Fraud Analytics

Deliverable

production ML models | product features | dashboards & analysis | client delivery | infrastructure

Required skills

Team leadership and management, strategic planning, SQL, Python, domain fluency in scaled abuse and fraud patterns, experience tracking threat actors across dark web environments, familiarity with large language model exploitation, process design from scratch.

Preferred skills

Experience at major tech platforms in trust and safety, background in financial crime investigation, experience with U.S. government stakeholders, track record of managing senior technical specialists, fluency in Mandarin Chinese or Russian, Active Top Secret security clearance.

Technologies

SQL, Python, Large Language Models, Dark Web tools, Proxy ecosystems.

Responsibilities

Own strategy and priorities for the mission area, hire and develop technical investigators, design lanes between management and senior ICs, lead complex investigations into model distillation and fraud networks, redesign triage for high-volume detection pipelines, expand coverage into fraud and scams, manage external engagement with U.S. government partners, convert findings into product mitigations, define and report team metrics.

Seniority

Senior, hands-on IC with management responsibilities

Sourced via greenhouse · Listed on CareerPlan, which tracks 70,000+ jobs from 20+ sources.
Apply on Greenhouse ↗