AI Safety Red Team Engineer | Equity & Flexible PTO

White Circle; Colombia; Remote; COP 120,000,000 - 180,000,000; Full time

Benefits offered by this job

Competitive compensation package

Equity

Flexible Time Off

Language lessons

Learning and development support

Team off-sites

Job summary

White Circle is seeking an AI Red Team Engineer to break LLM-powered systems responsibly, automate attacks, and turn findings into clear evidence for customer demos, security reviews, and sales conversations. You'll own hands-on adversarial testing end to end: find the failure, prove it, script it, and write it up.

You'll build lightweight Python tooling, maintain an internal attack library, and convert model failures into regression tests and product requirements to secure AI systems at scale.

Qualifications

• Experience in QA automation and security testing of APIs, web apps, and SaaS.
• Strong Python scripting skills for building attack tooling and automation.
• Hands-on with LLM prompts, system instructions, RAG, and tool calls.
• Background in vulnerability research, bug bounty, or AppSec is a plus.

Responsibilities

• Red-team LLM-powered systems and identify jailbreaks, prompt injections, data leakage, and unsafe outputs.
• Develop lightweight Python tooling to automate attacks and generate reports.
• Build and maintain an internal attack library with test cases and regression suites.
• Turn model failures into clear regression tests and product requirements.
• Produce credible evidence for customer demos, security reviews, and GTM efforts.

Skills

QA automation

API security testing

Python scripting

LLMs / prompts

Bug bounty experience

Adversarial testing

Tools

Burp Suite

Postman

Playwright

pytest

LangChain

Job description

White Circle is seeking an AI Red Team Engineer to break LLM-powered systems responsibly, automate attacks, and turn findings into clear evidence for customer demos, security reviews, and sales conversations. You'll own hands-on adversarial testing end to end: find the failure, prove it, script it, and write it up.

You'll build lightweight Python tooling, maintain an internal attack library, and convert model failures into regression tests and product requirements to secure AI systems at scale.