AI Red Team Engineer

Role overview Test AI-powered products through responsible, hands-on adversarial work, then turn the results into clear evidence teams can use to improve security and reliability. You’ll investigate failures across chatbots, copilots, retrieval systems, agents, and APIs; automate repeatable tests; and communicate findings for product work as well as customer-facing reviews and demonstrations.

Responsibilities - Probe AI applications for jailbreaks, prompt injection, data or instruction leakage, unsafe responses, policy bypass, tool misuse, excessive autonomy, resource abuse, and business-logic weaknesses. - Write Python scripts to run attack prompts against model APIs, gather and score responses, and generate repeatable results. - Maintain a reusable library of scenarios, prompts, test cases, scoring rubrics, regression checks, and demonstrations. - Document failures with reproduction steps, impact, severity, and practical remediation guidance. - Turn confirmed weaknesses into regression tests and product requirements, and incorporate useful new testing techniques. - Prepare credible technical evidence for product demonstrations, security reviews, and customer discussions.

Requirements - Experience in QA automation, application security, API or penetration testing, or bug bounty work. - Strong Python scripting skills and experience testing APIs, web applications, backends, or SaaS products. - Practical familiarity with language models and adversarial testing of AI systems. - Ability to work from incomplete requirements and explain technical findings clearly. - Sound ethical judgement: conduct testing within authorised scope and applicable law, with the aim of making systems safer.

Nice to have - Experience with Burp Suite, Postman, Playwright, or pytest. - Familiarity with automated AI red-teaming, retrieval systems, agents, tool-calling, or evaluation methods. - Knowledge of AI or web security frameworks and taxonomies. - Experience writing customer-facing security reports or building evaluation pipelines, regression suites, dashboards, and CI security checks. - Background in trust and safety, abuse prevention, fraud, moderation, platform security, CTFs, or responsible disclosure.

Benefits and work setup - Compensation includes equity and flexible time off. - Language-learning support for English or French, plus learning and development funding. - Hardware, subscriptions, and other tools are provided; team off-sites take place twice a year.