Principal Expert – Security of AI, Models, and Agents

About Quantum Beyond

Quantum Beyond advises companies and organizations on their strategic decisions related to technology, artificial intelligence, cybersecurity, digital sovereignty, and organizational transformation. Our experts work independently alongside leadership teams and strengthen internal teams as well as their partners.

Your mission

You will assess and strengthen the security of the artificial intelligence models, agents, and architectures used or developed by Quantum Beyond’s clients. You will work on the risks specific to generative AI, agentic systems, RAG architectures, and the interactions between models, data, tools, and users.

You will also contribute to securing Quantum Beyond’s products and research work.

Key responsibilities

• Perform threat modeling specific to AI systems.
• Assess the security of models, agents, data, and RAG architectures.
• Analyze the risks of direct or indirect prompt injection.
• Assess the potential for data exfiltration, poisoning, or manipulation.
• Examine the permissions granted to AI agents.
• Secure agents’ access to tools, APIs, systems, and databases.
• Define human oversight and authorization mechanisms.
• Carry out or lead red teaming exercises on AI systems.
• Assess the risks associated with external models, libraries, and vendors.
• Design mechanisms for monitoring behavior.
• Define detection, containment, and shutdown scenarios.
• Take part in designing private and sovereign AI architectures.
• Validate the corrective measures implemented by the teams.
• Produce technical and executive reports.
• Train developers, architects, and security leaders.
• Contribute to Quantum Beyond’s methods, tools, and research work.

Profile sought

• Minimum of 8 years of experience in cybersecurity, AI, or secure software engineering.
• Hands-on experience in the security of artificial intelligence systems.
• In-depth knowledge of LLMs, agents, RAG architectures, and APIs.
• Command of threat modeling and security testing.
• Understanding of the risks related to data, identities, and supply chains.
• Ability to work hands-on in complex technical environments.
• Strong analytical rigor and excellent professional judgment.
• Ability to explain risks clearly to executives.
• Professional French and English.

Assets

• Experience in red teaming of AI systems.
• Knowledge of security frameworks specific to AI.
• Experience in DevSecOps, MLOps, or LLMOps.
• Experience in application, cloud, or API security.
• Relevant scientific, technical, or open source contributions.
• Master’s degree or doctorate in a related field.

Conditions

Location (on site and remote), compensation based on experience, benefits and working arrangements: to be discussed or determined...