Trust & Safety Engineer — AI-Driven Threat Detection

Vercel; New York (NY); On-site; USD 196,000 - 294,000; Full time

Benefits offered by this job

Competitive pay

Equity

Healthcare

Mentorship & events

Flexible PTO

WFH budget

Job summary

Vercel is seeking an experienced security-focused engineer to join our Trust & Safety team in New York. You will analyze threat actor behavior, build scalable detection systems, and drive production-ready abuse prevention features.

You will collaborate with security, product and infra teams, own end-to-end projects from research to deployment, and leverage LLM-driven techniques to stay ahead of evolving abuse.

Qualifications

• 5+ years of experience in abuse prevention, trust & safety, or security engineering.
• Strong programming skills in JavaScript/TypeScript and Python.
• Hands-on experience with LLMs, including prompt engineering, evaluation, and fine-tuning.
• Experience owning systems end-to-end, from initial research through production deployment and iteration.

Responsibilities

• Analyze threat actor behavior and abuse patterns across large datasets to inform detection logic, policy decisions, and scalable mitigations.
• Research, prototype, and implement state-of-the-art LLM-driven techniques for abuse detection, classification, and prevention.
• Design and develop production-ready systems that detect and disrupt abusive behavior across the Vercel platform.
• Build, maintain, and fine-tune internal tools, datasets, and evaluation pipelines for trust and safety workflows.
• Collaborate closely with security, product, and infrastructure teams to bring research ideas into stable, scalable production systems.

Skills

JavaScript/TypeScript

Python

LLMs

AWS

Threat actor behavior

End-to-end ownership

Job description

Vercel is seeking an experienced security-focused engineer to join our Trust & Safety team in New York. You will analyze threat actor behavior, build scalable detection systems, and drive production-ready abuse prevention features.

You will collaborate with security, product and infra teams, own end-to-end projects from research to deployment, and leverage LLM-driven techniques to stay ahead of evolving abuse.