Senior AI Security Researcher – Red Team & LLM Safety

Snyk Ltd.; United States; Teletrabalho; USD 150 000 - 210 000; Tempo integral

Resumo da oferta

Snyk Ltd. is seeking a Research Engineer to probe AI and autonomous agent systems, develop attack tooling, harnesses, and reproducible evaluations. The role drives measurable improvements in AI security and strengthens Snyk's offensive security offerings.

You will focus on AI red teaming, offensive security testing, LLM fundamentals, harness engineering, and experimental design, using Python, APIs, containers, and browser automation to build maintainable research tools.

Qualificações

• Experience in AI red teaming, prompt injections, jailbreaks, and data leakage.
• Strong Python skills and experience with APIs and tooling.
• Background in offensive security testing of web/API workloads.
• Familiarity with LLM behavior, context management and tool calling.
• Experience building reproducible research tooling and harnesses.

Responsabilidades

• Research and exploit weaknesses in LLM applications and autonomous agents.
• Build attack tooling, agent harnesses, and reproducible evaluations.
• Improve Snyk's offensive security product offering through rigorous assessments.
• Collaborate with engineering to integrate findings into products.

Conhecimentos

AI red teaming

Offensive security

LLM fundamentals

Harness engineering

Agent security

Evals and benchmarks

Experimental design

Research engineering

Python

APIs

Containers

Browser automation

Ferramentas

Python

APIs

Containers

Browser automation

Descrição da oferta de emprego

Snyk Ltd. is seeking a Research Engineer to probe AI and autonomous agent systems, develop attack tooling, harnesses, and reproducible evaluations. The role drives measurable improvements in AI security and strengthens Snyk's offensive security offerings.

You will focus on AI red teaming, offensive security testing, LLM fundamentals, harness engineering, and experimental design, using Python, APIs, containers, and browser automation to build maintainable research tools.