Job Specifications & Details
Role Overview
Join a team where AI agents are an everyday working tool. This role offers real production tasks, not just theoretical security. You will gain direct access to top LLMs (Claude, GPT, open-source) and budgets for experiments. Remote or hybrid work format is available.
Key Responsibilities
• Design security architecture for agent systems: sandboxes for tool execution, circuit isolation, limits, kill switch• Protection against prompt injection, data exfiltration, jailbreak at the architecture level, filters, and output validation• Conduct Red Team exercises against proprietary agents• Review agent pipelines from the LLM engineering team through to production deployment• Control what is sent to external LLM providers
Requirements & Qualifications
• 3-5 years of experience in DevSecOps / AppSec / Security Engineering• Proficient in Python, experience with Docker, Linux, CI/CD• Practical experience with Threat modeling• Experience with Secrets management (Vault or similar), IAM, sandbox environments• Understanding of how LLMs and agent systems (LangChain / LangGraph / MCP / Claude Code) work• Readiness to design and build agents independently, not just review existing code