ExtraMint AI is building a team for a 2-week project in AI safety evaluation.
You'll design Dockerized scenarios that test whether AI coding agents stay within policy while completing realistic engineering tasks.

What you'll do
Build self-contained environments: realistic repos, tasks and test fixtures
Write deterministic graders and reference solutions
Red-team AI agents to find where they cross defined safety policies
Iterate quick: one scenario plus several meaningful variants

You should have
A security background (pentesting, red team, CTFs, or SOC/IR)
Strong Python and Linux, plus Docker
An adversarial mindset and attention to detail
Prior LLM or prompt red-teaming experience (a big plus)

Payment: $10 per accepted RLE.

Trial task: before onboarding, you'll complete a sample package to confirm fit. The sample is unpaid unless accepted, in which case it converts to paid.

Engagement: remote, about 2 weeks, starting soon.
To apply, send your CV/LinkedIn and one line on your most relevant security project to [your business email].
cybersecurity #AIsafety #redteaming #hiring

📌 Cybersecurity Engineers for AI Safety Evaluation (India)
🏢 ExtraMint AI
📍 India