AI Red Team Simulation Engineer

Company: Saffroot Technologies LLP

Role summary: Key Responsibilities - Design red-team simulations for LLM applications and agentic workflows. - Create adversarial prompts to uncover jailbreaks, policy failures, hallucinations, and tool-misuse risks. - Build evaluation workflows for safety regressions, failure categorization, and mitigation verification. - Investigate unsafe model behavior and convert findings into repeatable test cases. - Develop Python tooling for attack replay, failure triage, and safety-reporting automation.

Location: Anywhere in India

Employment type: Fulltime

Experience: 3-6

Salary / budget: ₹20-30 LPA

Notice period: 15 days

Skills: [{"name":"AI safety evaluation","weight":0.4},{"name":"prompt attacks","weight":0.4},{"name":"failure analysis","weight":0.2}]

Date posted: 2026-04-30

Why this role matters

Key Responsibilities - Design red-team simulations for LLM applications and agentic workflows. - Create adversarial prompts to uncover jailbreaks, policy failures, hallucinations, and tool-misuse risks. - Build evaluation workflows for safety regressions, failure categorization, and mitigation verification. - Investigate unsafe model behavior and convert findings into repeatable test cases. - Develop Python tooling for attack replay, failure triage, and safety-reporting automation.

Related job pages

Job description

Key Responsibilities - Design red-team simulations for LLM applications and agentic workflows. - Create adversarial prompts to uncover jailbreaks, policy failures, hallucinations, and tool-misuse risks. - Build evaluation workflows for safety regressions, failure categorization, and mitigation verification. - Investigate unsafe model behavior and convert findings into repeatable test cases. - Develop Python tooling for attack replay, failure triage, and safety-reporting automation.

Open candidate page and apply