Latest AI JobsBrowse Jobs
Handshake logo

AI Red Teamer (Remote)

HandshakeRemote (USA)HAI Delivery Ops
Apply on company site
Salary
$32 – $95 per hour
Job type
Full-time
Work mode
Remote
Experience
Mid
Posted
5 Oct 2026
Status
Open · checked 7d ago

About this role

The AI Red Teamer position is a remote role based in the USA that involves stress-testing large language models to uncover safety and security vulnerabilities. This position suits individuals with strong ethical judgment, adversarial problem-solving skills, and a background in creative or analytical testing who want to support AI safety for leading research labs.

Role Overview

As an AI Red Teamer, the professional evaluates large language models by intentionally attempting to break system guardrails. The worker designs adversarial prompts to expose potential weaknesses across multiple risk domains, including content safety, cybersecurity, and regulatory compliance. The employment type is W-2 and hourly, operating for forty hours per week through TCWGlobal, with a location restriction excluding residents of California.

Key Responsibilities

Build creative multi-turn prompts and scenarios to test AI defenses across various risk categories.
Find methods around safety restrictions using evasion, jailbreak, and prompt injection techniques.
Investigate edge cases to trigger disallowed, incorrect, or harmful outputs.
Score and evaluate model responses according to structured severity rubrics and harm taxonomies.
Record experiments thoroughly, detailing methodology, rationale, and findings.
Assess and improve adversarial prompts created by fellow team members.
Participate in calibration exercises and harm taxonomy development.
Partner with researchers, data scientists, and engineers to communicate findings and improve defenses.
Handle potentially disturbing content regularly.
Keep track of current attack methods, jailbreaks, and evolving model behaviors.

Requirements

Demonstrated hands-on experience utilizing multiple large language models such as Claude, ChatGPT, Gemini, and open-source models.
Intuition for designing adversarial prompts, along with familiarity regarding evasion or jailbreak techniques.
Adversarial and creative problem-solving capabilities.
Clear written communication skills.
Strong ethical judgment to separate personal values from adversarial thinking.
Self-directed and collaborative work style within feedback-heavy environments.
Persistence, curiosity, and comfort handling frequent experimental failure.
Ability to engage professionally and sustainably with harmful content.

Nice to Have

Knowledge of scripting languages like Python.
Experience using evaluation tooling or large language model application programming interfaces.
Comfort performing rubric-based scoring and structured data annotation.
Background experience in security research, quality assurance, content moderation, or trust and safety.
Subject matter expertise in high-risk areas including law, finance, chemistry, biology, medicine, or cybersecurity.

Skills and Tools

ChatGPT
Claude
Gemini
Open-source models
Python
Large language model APIs and evaluation tooling
Adversarial prompting and jailbreak techniques
Rubric-based scoring and structured data annotation

Salary and Benefits

Compensation ranges from $32 to $95 per hour.
Position is W-2, hourly, and employed through TCWGlobal at forty hours per week.

About the Company

Handshake collaborates with leading artificial intelligence research labs to enhance model safety and robustness. The organization conducts red teaming operations to discover vulnerabilities prior to user release, supporting the responsible advancement of frontier artificial intelligence systems.

How to Apply

Candidates should apply on the official company careers page.

Skills and tech stack

ChatGPTClaudeGeminiPythonLLM APIsLLMLLMs

Questions candidates ask

What is the location requirement for the AI Red Teamer role?

The position is remote within the USA, but candidates residing in California are ineligible for hire.

What is the employment type and compensation?

It is a W-2, hourly role at forty hours per week, employed through TCWGlobal, paying between $32 and $95 per hour.

What kind of content will I be exposed to in this job?

The role involves regular and deliberate exposure to harmful content, including violence, self-harm, hate speech, sexually explicit material, and child safety scenarios as part of adversarial testing.

What are the core technical expectations for applicants?

Applicants need strong hands-on experience using multiple large language models like ChatGPT, Claude, Gemini, and open-source models, alongside an intuition for crafting adversarial prompts.

How should I submit my application?

Interested applicants should apply on the official company careers page.

Ready to apply?

The application is on the official Handshake careers page.

Apply on company site