Experteer Overview In this role you drive safe, trustworthy agentic AI for pharmaceutical Ru0026amp;D within Johnson u0026amp; Johnson’s Data, Data Science u0026amp; AI group in Madrid or Barcelona. You define provable controls, adversarial assurance, and safety-native architectures to ensure AI systems are safe by design for regulated workflows. You will collaborate with cross-functional teams to translate safety requirements into scalable controls and prototypes, shaping the GenAI Platform and contributing to scientific and regulatory proof. This is an impactful, hands-on role at the frontier of AI safety in life sciences. Compensaciones / Ventajas • Design machine-readable control policies governing agent actions, tool use, data access, and information flow across agentic workflows. • Implement deterministic policy enforcement with auditable decisions and defined human-approval paths. • Enable domain and system owners to author, test, and maintain controls via accessible policy interfaces. • Develop continuous red-teaming methods for agentic AI,



integrating threat models with Ru0026D failure modes. • Embed adversarial evaluation into the GenAI Platform to test models, agents, tools, and end-to-end workflows. • Define evaluation protocols and evidence thresholds to distinguish safety properties from unverified claims. • Research and prototype safety-native architectures that constrain agent behavior with layered controls. • Evaluate pre-training data interventions and post-training methods for regulated scientific use cases. • Translate scientific, quality, privacy, and regulatory requirements into testable system specifications for models, agents, tools, and runtime controls. • Define accountability measures and ensure control decisions are traceable and reconstructable as audit evidence. • Partner with Ru0026D scientists, platform engineering, security, privacy, legal, and quality teams; guide a small cross-functional team and represent work externally through

📌 Senior Scientist (Comunidad de Madrid)
🏢 Johnson u0026 Johnson
📍 Comunidad de Madrid