AI Red Team Intern: Break & Harden Safe Models

Realm Labs; Sunnyvale (CA); On-site; USD 34,000 - 55,000; Full time

Benefits offered by this job

Market aligned intern compensation -,u

Job summary

Realm Labs in Sunnyvale, CA is seeking an Intern for AI Red Teaming in Fall 2026. This in-office, full-time role focuses on breaking systems and models, turning findings into reproducible attacks, and documenting why they work.

You will engage with adversarial ML methods and agentic systems, with a goal to publish a technical report and improve the evaluation suite. Ideal candidates have hands-on attack experience, ability to implement research papers, and familiarity with ML tools and security

Qualifications

• Hands-on experience attacking or stress-testing models.
• Read a paper and implement its attack.
• (nice to have) Offensive security background outside ML: CTFs, vulnerability research.
• (nice to have) Familiarity with agentic systems and their attack surface.

Responsibilities

• You will try to break the systems we build and the models we protect, and turn what you find into something the team can act on: a reproducible attack, an evaluation that catches it, and a written account of why it works.
• Expect mix of eliciting unsafe behaviour from aligned LLMs and multi-modal models; prompt injection and tool-use abuse against agentic systems; automating attack generation and evaluation; and measuring guardrails under pressure.
• Aim for a paper or public technical report out of every internship, plus attacks that stay in our evaluation suite after you leave.

Skills

Adversarial ML

Red Teaming

Python

Pytorch

LLM safety

Tools

AWS

GCP

Jupyter

Job description

Realm Labs in Sunnyvale, CA is seeking an Intern for AI Red Teaming in Fall 2026. This in-office, full-time role focuses on breaking systems and models, turning findings into reproducible attacks, and documenting why they work.

You will engage with adversarial ML methods and agentic systems, with a goal to publish a technical report and improve the evaluation suite. Ideal candidates have hands-on attack experience, ability to implement research papers, and familiarity with ML tools and security