OpenTrain AI
last yearAI Red Team Engineer for LLM Security
The work You will design and run adversarial security tests against large language models, AI agents, and retrieval-augmented generation pipelines. The work focuses on finding weaknesses, measuring model behavior, and documenting clear ways