AI security engineers: find and fix flaws in LLM apps

We're staffing security engineers to build and analyse vulnerable LLM agent and RAG code for AI training and evaluation work. It's remote. This takes about a minute.

• The work: build small, runnable LLM agent and RAG codebases that contain realistic security flaws (sensitive information disclosure, excessive agency) plus fixed versions, then trace, annotate and test each finding.
• You need security experience (application security, secure code review, penetration testing or security-focused engineering) and Python or TypeScript. Hands-on work with LLM agents, RAG or MCP is a strong plus.
• Full time (40 hours) or part time (20 hours) a week for about 4 weeks, with 4 hours of your day overlapping US Pacific time.
• Open in Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Indonesia, Nigeria, Turkey and the United States.