Accenture, recognized as a Great Place To Work®, is a leading global professional services company that helps the world’s largest businesses, public administrations and other organizations develop their digital core, optimize operations, accelerate revenue growth and improve citizen services, creating tangible value at speed and scale. We are seeking an RAI Red Teaming / AI Reliability specialist who can safely 'break' AI systems, including agentic AI. You will transform emerging threats into concrete adversarial tests, translating findings into practical solutions and collaborating with research, engineering, product, legal and compliance teams to harden LLMs, RAG systems and AI agents. You are passionate about hands-on experimentation, rapid iteration and AI. Responsibilities: * Plan and execute end-to-end RAI Red Teaming evaluations: scoping, attack hypotheses, execution, evidence collection and debriefs * Emulate real-world threats: prompt injection, jailbreaks, data exfiltration, model hijacking, abuse of agents/tools, multi-agent collusion, memory poisoning * Build and maintain adversarial prompt datasets and evaluation suites * Develop automated evaluation pipelines (CI/CD) to track attack success rates, severity, coverage, regressions and agent security KPIs * Conduct AI-specific threat modeling (LLM/ML attack catalogs, OWASP Top 10 for LLMs, agentic threat trees) and propose prioritized mitigations * Validate guardrails, policies, content moderation, sandboxing, tool/API permissioning and human-in-the-loop controls * Perform ethical testing and fairness/bias audits; evaluate hallucinations, safety, privacy, toxicity and IP leakage * Co-create RAI Red Teaming training programs and agent security best practices Suggested education: * Computer Engineering, Mathematics, Physics, Telecommunications, Cybersecurity, Artificial Intelligence, Statistics or related fields * Master's or PhD in technical areas related to AI, ML, NLP or cybersecurity * Relevant certifications: OWASP GenAI Red Teaming, NIST AI RMF, ISO/IEC 42001, CSA Agentic AI Red Teaming Languages: * Fluency in Spanish, both spoken and written, is essential to collaborate with local teams and understand technical and legal documentation * Strong English skills are desirable for interaction with international teams, participation in trainings and access to specialized documentation Valued experience: * Candidates with at least 3 years of red teaming experience (with at least 1 year in generative AI) and profiles with 5 years of experience, including technical or team leadership in AI security projects, are preferred * Prior experience with agentic systems will be valued This position is based in Madrid or Barcelona, with potential consideration for locations such as Málaga or Zaragoza, under a hybrid work model involving some days from home and others in the office, where you can create valuable synergies with your team. Residency in Spain and a valid work permit in Spain are required. At Accenture, a company recognized as a Great Place To Work®, we support a hybrid work model that, thanks to technology and our facilities, enables us to maintain the essential human connection when working with our teams and clients. This connection allows us to preserve our culture of inclusion and diversity and be, according to Refinitiv, the most diverse company in the world. Additionally, we offer benefits such as: * Medical, life and accident insurance * Medical services and wellness programs * Flexible compensation program and stock purchase plan * Flexibility programs (flexible hours, free days, vacations...) * Individualized training itinerary * Sustainability programs and Accenture Foundation * Employee networks for diversity * Other benefits: Bankinter office with special conditions and profit sharing